Participating in a Story
Participating in a Story
EXPLORING AUDIENCE COGNITION
PIERRE GANDER
Lund University Cognitive Studies 119
Copyright © Pierre Gander 2005. All rights reserved.
For further information, corrections, and updates, see the author’s
web page at http://www.pierregander.com
Cover: Detail from Henna by Afsaneh Monemi. Photograph by
Afsaneh Monemi.
Printed in Sweden by Arkitektkopia, Göteborg 2005.
ISBN 91-628-6417-3
ISSN 1101-8453
ISRN LUHFDA/HFKO-1015-SE
Lund University Cognitive Studies 119
To everyone who
participates in my story
CONTENTS
Acknowledgements
CHAPTER
1
ix
Introduction
1
•
PART I
: THEORETICAL FRAMEWORK
CHAPTER
2
Stories
21
CHAPTER
3
Fiction
33
CHAPTER
4
Participatory stories
47
•
PART II
: EMPIRICAL STUDIES
CHAPTER
CHAPTER
CHAPTER
CHAPTER
5
6
7
A method for generating data
for exploring off-line cognition
Long-term memory
representations of spatiality
93
135
Exploring memory qualities using
the reality monitoring framework
173
8
Perspective on actions and events
199
9
Conclusions and future research
•
CHAPTER
References
Appendix
255
265
241
ACKNOWLEDGEMENTS
:HVWRI+RXVH
<RXDUHLQDQRSHQILHOGZHVWRIDELJZKLWH
KRXVHZLWKDERDUGHGIURQWGRRU
7KHUHLVDVPDOOPDLOER[KHUH
!2SHQPDLOER[
2SHQLQJWKHPDLOER[UHYHDOVDGLVVHUWDWLRQ
!5HDGGLVVHUWDWLRQ
ACKNOWLEDGEMENTS
During the years of working with this book, which is also my doctoral dissertation, my ideas and skills have formed in the meeting
with numerous people, some of whom I may not even remember,
at university departments, conferences, and elsewhere. Academically I have spent time at the Department of Computer Science,
University of Skövde, at the Department of Cognitive Science
(L U C S), Lund University, at the Department of Linguistics and the
S S K K I I Center for Cognitive Science, both at Göteborg University.
It is the people at these places who have shaped me the most. I also
had the pleasure of attending a couple of highly interesting summer
schools arranged by the Graduate School in Language and Communication, University of Southern Denmark, where Johannes Wagner played a leading organising role.
More directly related to the work with the dissertation, I wish to
thank my supervisor Peter Gärdenfors for giving me freedom and
for being so supportive and my auxiliary supervisor Sverker Sikström for being so critical. I am grateful to Anki Månsson, Kenneth
Wilhelmsson, Hans Vappula, Christer Lörnemark, Håkan Edqvist,
and Johanna Nord for help with transcriptions and codings. I acknowledge the Department of Linguistics at Göteborg University
for letting me use their facilities. Jim Lodge assisted me with video
technology. I am thankful to the participants in my study for their
cheerful cooperation in the sometimes lengthy sessions. For providing valuable comments and proofreading parts of my dissertation
ix
x
Participating in a story
draft, I would like to thank Mikael Jensen, Susanna Radovic, Leonard Ngaosuvan, Johan Språng, Ingvar Lind, Kenneth Wilhelmsson,
Sylvie Porhiel, Dominique Klingler, Anna Vallin, and Alan Crozier
(any remaining errors are of course entirely my own). Ingvar Lind
also assisted during the data collection and helped with the front
page graphics. I am grateful to Afsaneh Monemi for the front page
art and photograph. A special thank you to Nataliya Berbyuk, without whose moral support this dissertation might never have been
completed—‘It’s finished!’
1
CHAPTER 1
Introduction
H
U M A N C O G N I T I O N —the
processes by which we acquire and
use knowledge—enables us not only to act appropriately in
the world, but also opens up doors to imaginary worlds. We can
think about possible ways to solve a problem. We are captured by
the fate of a character in a fictional story which is told to us. We can
pretend that things in the world are other than what they actually
are, and we can pretend to do things other than what we actually
do. This book explores parts of the imaginative powers of the human mind, namely the cognition of taking part in a fictional story.
The ability to conceive of perception and action as relating to something other than the immediate ‘real world’ is a general human cognitive ability. As shown by studies of children’s pretence play, the
ability spontaneously develops early in childhood. However, this
phenomenon is not restricted to children; it is present also in adults,
which is the study population in this book.
Traditionally, stories,1 such as fairytales, printed books and cinema, do not let the audience influence the sequence of events in the
story. The interpretation of these stories can vary among people
and after multiple readings, and they can create a sense of involvement, but the audience is not a true participant. In contrast, in new
media, such as computer games and other interactive multimedia
(and also in non-computerised situations such as the older tradition
of role-playing, and in children’s pretence), the audience can become participants in the story. In this book, these stories are called
The term ‘story’ is used as a synonym to the term ‘narrative’ in this book. For a
discussion, see Chapter 2.
1
1
2
Participating in a story
participatory stories.2 Participatory stories allow the audience to
influence which events take place in the story. A challenge for literary studies, film studies, narratology, media theory, computer game
theory, and other similar fields which deal with the concept of
story, is then to account for the differences between nonparticipatory stories and participatory stories. The argument in this
book is that no such viable theory exists.
The main goal of this book is to argue that cognition is the best
place in which to look for differences between participatory stories
and non-participatory stories. The contribution can be divided into
two parts. First, the book provides a theoretical framework for participatory stories. A cognitive perspective on stories and fiction is
adopted. It is concluded that earlier work fails at successfully capturing the distinction between non-participatory and participatory
stories. In order to uphold this distinction, it is argued, a definition
of participatory stories must refer to cognition. The second type of
contribution is findings from empirical investigations of cognition in
actual participatory story situations, connecting the theoretical
framework to authentic situations where people are involved in a
computerised participatory story. Although the book deals
specifically with cognition in connection to new technology, its
framework is broader. It is the general cognitive ability of imagination (with its philosophical issues) that forms the basis for the book;
more specifically, having beliefs about fictional entities and acting
according to these beliefs. In the book, the theoretical framework
and the empirical studies are constructed with a firm belief in interdisciplinarity. Research is drawn together from cognitive psychology, linguistics, developmental psychology, reading studies, literary
studies, film studies, artificial intelligence, among others.
In this chapter, the purpose, the object of study, the scientific
framework, the delimitations, and the contributions of the book
The term ‘participatory stories’ is introduced in this book and is further discussed
and justified in Chapter 4, where also a definition of the concept of participatory
stories is given. The meaning of the term ‘participatory’ is similar to the meaning of
the term ‘interactive’. The phenomena go under other names in earlier research, such
as interactive stories, interactive narrative, and digital narrative.
2
Introduction
3
will be discussed. But first, the concept of participatory stories will
be clarified through an example of an actual participatory story.
1.1
PARTICIPATORY STORIES
The very idea of stories is changing. Conceptions of stories, thousands of years old, originating with Aristotle (350 B C /1999), are
being challenged by the technology of the late twentieth century. A
story may no longer need a fixed beginning, middle, and end, as
were some of Aristotle’s defining features. Neither did he mention
the possibility that the audience may influence the sequence of
events in the story (in this book called participation), which is possible with computer technology. An example is computer games in
which the user takes the role of a character in an unfolding story,
influenced by what actions are taken (however, not all computer
games are considered participatory stories; only those framed in a
story context—see Chapter 4). But story participation is not
uniquely a new thing and is not associated only with computers.
Since the 1970s, people have been engaged in role-playing games
(e.g., Dungeons and Dragons, and others) where a story emerges as a
result of the participants’ actions. More loosely structured roleplaying has been exercised throughout history, and children’s pretence play is surely as old as mankind itself. These forms of participatory stories are all expressions of a feature of the human mind:
the ability to have pretence beliefs and act according to those beliefs, without confusing them with actual beliefs about the world.
Chapter 4 is devoted to a discussion and definition of participatory stories. Let us here just briefly consider an example of a participatory story. The example is a participatory story called Anchorhead, which is a type of computer game called interactive fiction
(see, e.g., Costanzo, 1986; Aarseth, 1997; Murray, 1997; Wilhelmsson, 2001) (a full classification of participatory stories is presented in Chapter 4). This participatory story was used as stimuli in
the empirical studies in Chapters 6–8. Anchorhead has a simulated
world containing places, objects, and characters. Interaction with
4
Participating in a story
Anchorhead is purely textual. The computer presents the current
situation in text on the screen and the human user—the player—
types commands to the system in near-natural language. The commands are carried out by the player character in the story world,
such as moving around, picking up and manipulating objects, and
interacting with other characters in the story. After each command,
the simulated world is updated and the player is again asked to enter
a command. Below is a transcript from Anchorhead, taken near the
start of the story. The input from the player is shown after the ‘>’
sign. Note that the system addresses the player as ‘you’. (A longer
transcript is available in the appendix.)
Outside the Real Estate Office
A grim little cul-de-sac, tucked away in a corner of the claustrophobic
tangle of narrow, twisting avenues that largely constitute the older
portion of Anchorhead. Like most of the streets in this city, it is ancient, shadowy, and leads essentially nowhere. The lane ends here at
the real estate agent’s office, which lies to the east, and winds its way
back toward the center of town to the west. A narrow, garbagechoked alley opens to the southeast.
>look at alley
The alley lies southeast, and appears to lead around the side of the
real-estate office.
>look at office
The door has a glass front with the name of the real estate company—
Benson & Brackhurst—stenciled across it. The blinds are drawn, the
lights are off inside, and no one appears to be home. Odd, since the
agent knew you were coming today.
>look at street
The cobblestones are slick from the rain and worn with many centuries’ passage.
>knock on door
You rap on the glass sharply, peering through it into the dark room
inside. Nobody answers. Strange; you just talked to the real estate
agent—Miss Benson, you think it was—yesterday. She was going to
meet you here.
Introduction
5
>open door
It seems to be locked.
>inventory
You are wearing your wedding ring, your trenchcoat and your
clothes; in addition, you have in your hand your umbrella.
The pockets of your trenchcoat are empty.
>look at umbrella
Olive green, with a hook-shaped handle. It’s a handy little thing,
when it rains. It’s currently open.
>look at ring
Staring at your wedding ring like that tends to make you sentimental.
1.1
THE PURPOSE OF THE BOOK
The book aims at supporting the following thesis: The difference
between traditional, non-participatory stories (such as books and films),
and participatory stories (such as story-based computer games) is one of
cognition. The general aim of the book is reached by a theoretical
and an empirical part, which are described in subsections below.
The book rests on three important foundations: cognitive theory, empirical studies, and authentic material and situations (Figure
1.1). First, cognition is considered an essential part of the account of
participatory stories (further discussed below). Second—in contrast
to many earlier studies of participatory stories—analysis, introspection, and speculation are not considered suitable methods for arriving at knowledge about participatory stories; empirical studies are
needed. Finally, there is a conviction that these empirical studies
should not be artificial, manipulated, and stripped of context, but
must have high ecologic validity by including authentic stimuli
(such as Anchorhead exemplified above) and authentic situations
(such as that the study situation of using Anchorhead is similar to a
real situation). Otherwise, the stimuli and situations may lead to
unnatural cognitive processing strategies and unnatural mental rep-
6
Participating in a story
resentations (Graesser, Millis, & Zwaan, 1997). The two latter
points are discussed in Chapter 5.
Let us now consider the two parts of the book in more detail: the
theoretical framework and the empirical studies, followed by a note
on how the book may have consequences for research on cognition
in general.
1.1.1 A theoretical framework for participatory stories
The theoretical framework, Part I of the book, comprises Chapters
2–4. The literature review and arguments are carried out in those
chapters and are only summarised here.
The question answered in the theoretical framework is: What are
participatory stories? In order to answer this question, the concepts
of story, fiction, and participation are defined and discussed. Chapter
2 introduces the concept of story as a cognitive construct, while
Chapter 3 does the same for the concept of fictionality. With these
foundations in place, Chapter 4 provides a classification system of
the concept of participatory stories and its related phenomena, end-
Figure 1.1. The three pillars of this book: cognitive theory, empirical studies, and
authentic material and situations.
Introduction
7
ing with a proposed definition. The main justification for the theoretical framework offered is that earlier work has not been successful in explaining the difference between participatory and nonparticipatory stories. The claim advanced here is that accounts that
use differences in media, structure, or physical actions are insufficient for upholding the distinction between the two classes.
Instead, the book makes a case for a cognitive perspective in the
study of participatory stories. To apply a cognitive perspective here
means to define the concepts in relation to human cognition. The
importance of a cognitive perspective can be framed in a weak and a
strong version:
Strong cognitive thesis: It is necessary to include cognitive aspects
in order to show what differentiates participatory stories from nonparticipatory stories (cognition is a necessary part of the distinguishing criteria in order to maintain a separation of the two classes).
Weak cognitive thesis: It is fruitful and enlightening to account for
cognitive mechanisms when characterising participatory stories.
Few other studies of participatory stories on a general level have
concerned themselves with cognition—a rare case is Wilhelmsson
(2001). Wilhelmsson embraced only the weak version of the cognitive thesis. It is argued that all other approaches to capturing the
difference between traditional stories and participatory stories fail,
and the only plausible solution is to adopt the strong version of the
cognitive thesis.
1.1.2 Empirical studies of cognition in participatory
stories
Part II of the book is the empirical studies, comprising Chapters
5–8. The details are given in those chapters, and only summarised
here.
Arriving at the conclusion that a cognitive perspective is both
fruitful and necessary in order to account for the difference between
participatory stories and non-participatory stories, as the result of
the theoretical framework, the next step is to ask more specific
questions concerning cognitive differences. Many earlier studies of
8
Participating in a story
what happens when playing computer games have been neither empirical nor concerned with cognition (e.g., Laurel, 1991; Aarseth,
1997; Murray, 1997; Ryan, 1997). The studies concerned with cognition of computer games have not been empirical (Wilhelmsson,
2001), and if they have been empirical, they have not been about
cognition (e.g., Johansson, 2000; Linderoth, 2004). The second purpose of this book is to provide findings from empirical studies of
cognition of participatory stories.
There are many methodological challenges when designing a
study which compares cognition of non-participatory and participatory stories.3 The potentially confounding factors are plentiful:
there is the computer interface, the participants’ computer skills and
varying skills with participatory stories. Most importantly, a participatory story is a different thing from a non-participatory story—
it does not have a fixed length, for instance. Time—how long a participant spends reading or using it—is not easily comparable across
non-participatory and participatory stories (a textual nonparticipatory story can be read at a steady tempo from beginning to
end, but a participatory story may halt, move slowly, or proceed
depending on the actions of the participant). For these reasons and
because little scientific knowledge has been collected about cognition of participatory stories, the empirical studies in this book are
exploratory in nature, rather than experimental and hypothesistesting. An effort in the empirical studies was to achieve a high ecological validity by including authentic stories and situations, as well
as selecting participants with sufficient experience of participatory
stories. The process of moving the phenomenon from its natural
context to a study context was carried out with careful consideration.
Every investigation, including an empirical one, is based upon philosophical assumptions. This dissertation rests on a set of philosophical assumptions called postpositivism (Guba & Lincoln, 1994): ontologically, the view is that there exists a single,
independent, external reality. Epistemologically, the view is that we can gain approximate knowledge of this reality. Methodologically, this should be done, at least
ultimately, using empirical methods aimed at objectivity (manifested practically as
intersubjectivity).
3
Introduction
9
An overview of the empirical studies is shown in Figure 1.2. A
single method for data collection was designed (Chapter 5), which
incorporated exposure of participants to events from five sources
with varying degrees of participation, fictionality, and authenticity:
a computer game, a printed short story, personally experienced
events, practical laboratory tasks, and a special, printed nonparticipatory version of the computer game. Afterwards, participants talked about events from these five source conditions in an
interview while being audio-visually recorded. The idea was to get
at how people think about participatory stories by analysing how
they talk about participatory stories. In this way, language (speech
and gesture) was used as a window to the mind. Computer logging
of interaction with a participatory story was also used to some extent. The collected data allowed multiple analyses of cognition from
three main viewpoints: spatial cognition (Chapter 6), memory
qualities of events and actions (Chapter 7), and what perspective
people adopt on events and actions (Chapter 8).
SOURCES
DATA COLLECTION
COMPUTER
GAME
ANALYSIS
SPATIAL
COGNITION
(CHAPTER 6)
DESCRIPTIONS
REPRESENTATIONS
SHORT STORY
INTERVIEWS
PERSONAL
EXPERIENCE
MEMORY
QUALITIES
(CHAPTER 7)
LABORATORY
TASKS
NON-PARTICIP.
VERSION
PERSPECTIVE
(CHAPTER 8)
Figure 1.2. Overview of empirical studies in this book.
SPEECH
GESTURE
10
Participating in a story
1.1.2.1 Spatial cognition (Chapter 6)
Spatial cognition, that is, how people acquire and use knowledge
about the physical surroundings, was investigated in the first study
(Chapter 6). Spatiality has been offered as a key feature of participatory stories (Murray, 1997). Studies of naturalistic reading (of nonparticipatory stories) have shown that readers do not spontaneously
form representations4 of the spatial layout of the story world in
long-term memory (Zwaan & van Oostendorp, 1993; Hakala, 1999).
Because of a navigational demand, unique to participatory stories, it
was predicted that the audience form spatial mental representations
in a way that readers of non-participatory stories do not. However,
a notion of cognition as basically situated may deny the presence of
such long-term memory representations and instead hold that people use cues in the current situation. The study investigated how
people talked about space from a participatory story and what this
tells us about their representations of space in long-term memory.
When analysing participants’ verbal descriptions of spatiality
from the participatory story, it was found that participants exclusively used a survey descriptive strategy (i.e., giving a description
from above) using an extrinsic frame of reference (e.g., north, south).
This was true for both the computer game and the nonparticipatory version. However, a marked difference was found
regarding spatial mental representations. Participants who played
the computer game revealed elaborate, relatively complete, accurate,
and integrated representations of spatiality of the participatory
story world in long-term memory. In contrast, participants who
read the non-participatory version of the computer game provided
little evidence to suggest that they formed spatial mental representations at all.
1.1.2.2 Memory qualities (Chapter 7)
Events that take place in a participatory story are fictional, but at
the same time they occur as a consequence of the audience’s actions.
Are memories of events from participatory stories different from
4
More is said on the issue of representations in Section 1.4 as well as in Chapter 6.
Introduction
11
memories of real events? 5 Are memories of events from participatory stories more like events read about in a short story? In the second study (Chapter 7), differences between memories of events
from the five sources are explored (see Figure 1.2) using the reality
monitoring framework (Johnson & Raye, 1981; Johnson, Foley,
Suengas, & Raye, 1988), which concerns memory qualities such as
richness of perceptual, spatial and temporal details.
Results revealed no differences in reality monitoring memory
qualities and were thus contrary to predictions from the reality
monitoring framework. The proposed interpretation is that the
reality monitoring framework is suited to explain differences between memories of events with external and internal origin, but not
suited to explain differences between memories with varying degree
of fictionality, as studied in this book.
1.1.2.3 Perspective on actions and events (Chapter 8)
Considering that the audience carries out actions in a participatory
story, from what perspective are events and actions seen? Are events
and actions seen from an ‘outside’ perspective, as if carried out by
someone else, similar to events from a fictional short story? Or are
events and actions seen from an ‘inside’ perspective, as if carried out
by oneself in the real world? Theoretical studies of computer games
suggest that the audience considers the agency in participatory stories as an extension of themselves (Wilhelmsson, 2001). Empirical
studies have revealed that the audience often uses the pronoun ‘I’
when talking about agency in computer games (Johansson, 2000;
Linderoth, 2004). In order to study perspective on actions and
events, two separate analyses were carried out of the interviews:
speech and gesture (Chapter 8).
The analysis of how perspective was revealed through hand gestures made by the participants while they were talking about events
suggests that participants sometimes viewed themselves as being
Real and reality are used in opposition to fictional—see Chapter 3 for a philosophical
discussion and Chapter 7 for an empirically related discussion in relation to the reality monitoring framework.
5
12
Participating in a story
outside the event, and sometimes as taking part inside the event.
The analysis of perspective in speech offers the clearest picture of
what perspectives were adopted when the participants were thinking and talking about events from the participatory story. Results
from analyses of perspective as manifested in speech revealed two
main groups. In the computer game, personal experience, and laboratory tasks, participants mainly used an inside perspective, but the
common use of the indefinite pronoun man (‘you’/‘one’), especially
when talking events and actions from the computer game, added
distance to the perspective. In the other two conditions, participants
mainly took an outside perspective. Thus, analysis of speech
showed that participants adopted a perspective on actions and
events of a participatory story that is similar to real, personally experienced events and not similar to fictional events read about in a
short story.
There was a frequent perspective switching between ‘I’ and the
indefinite pronoun man when participants talked about events from
the computer game. The most frequent reason for switching from
‘I’ to man was to talk about events in the game that would happen
to any player. In this way, the switch allowed the participants to be
more general. The most common reason for switching in the other
direction, from man to ‘I’, was to tell about specifics of what the
participant did in the computer game. In this way, the use of ‘I’ expresses that what is being described from the game is something
occurring because of an action—a conscious decision—on the part of
the player.
The results of the empirical studies have consequences for the
weak cognitive thesis. The weak cognitive thesis is substantiated by
findings concerned with spatial cognition and perspective on
memories of events and actions, but not by findings from phenomenal memory qualities. In other words, it is fruitful and
enlightening to talk about differences in cognition in relation to
spatial cognition and perspective on events and actions, but memory qualities is an area in which cognition does not seem to add to
the account of participatory stories. However, it should be pointed
out that no complete theory of the cognition involved when experi-
Introduction
13
encing participatory stories is presented in this book. Rather, the
empirical results of the book place constraints on what such a theory may look like.
1.1.3 Implications for theories of cognition
The investigations in the present book also have implications for
theories of cognition not concerned specifically with participatory
stories. Using participatory stories as a laboratory for studying the
general human cognitive ability of imagination, conclusions can be
compared against other theories of cognition, such as reality monitoring. Questions are also raised concerning the notion of ‘self’: Are
actions carried out in a participatory story thought of as having
happened to the self? If so, what is the difference, if any, between
these memories and autobiographical memories? The phenomenon
of participatory stories raises questions regarding the notion of the
self in research on memory.
1.2
COGNITION
The object of study in this book is cognition. The view of the nature of cognition and mental events has shifted historically, so the
assumptions on which the work in this book rests need to be discussed.
In this book, cognition is taken to mean the processing of information in the human mind.6 Cognition is seen in contrast to emotion and motivation, which are assumed to be analytically separate
Philosophically, this view is connected to functionalism (Block, 1980). However,
the work in this book places few constraints on the concept of consciousness. As the
concept is riddled with problems, it was considered wise to exclude it when it has no
explanatory role. Thus, the book adheres particularly to decompositional functionalism and computation-representation functionalism, and not metaphysical functionalism.
Further, the cognitive issues discussed in this book are mainly high-level issues. Thus,
arguments and empirical findings are consistent with both symbolic and connectionist
accounts of cognition, which are here regarded chiefly as a matter of low-level realisation.
6
14
Participating in a story
phenomena. Cognition involves processes and mental representations which a person is aware of, as well as those of which a person
is unaware. The former can usually be explicitly verbalised while
the latter cannot—they emerge in behaviour (including nonconscious parts of language use). Both are included as objects of
study in this book.
In this view, there is an assumption regarding generality of cognition. Although the content of cognitive processes and mental representations varies across individuals, the form and functioning of cognition are general and similar across individuals.7 This assumption
connects to the nomothetic/ideographic disjunction (Guba & Lincoln, 1994): What should be described: general laws or individuals’
specific traits? The present work takes a nomothetic approach, in
that it studies general processes, but allows for ideographic differences concerning the content of those processes.
A challenge to the view that cognition is information processing
in the mind is the notion of situated/distributed cognition, as advanced by, for example, Hutchins (1995). The idea is that cognition
does not primarily take part in the heads of people, but that cognition always involves the brain, body, and environment in which it
is situated. The argument is that once this perspective is taken there
is little need to postulate internal representations of the world. Why
not adopt a situated-cognition view of participatory stories? In
agreement with Hutchins (1995), an important assumption in the
present book is that cognition ought to be studied ‘in the wild’, in
the sense that what is of interest is cognition taking place in its
natural context. But this does not necessarily lead to a view of cognition as fundamentally distributed. In some domains, natural cognition may still be non-distributed and internal. For instance, consider the case where a person is reading a fictional novel. Here, there
is little of interest in terms of interaction with the environment. In
However, there may be differences between specific groups, such as between novices and experts, and healthy versus pathological individuals. The factor of culture is
not a concern in this book, and thus no assumptions regarding the universality of
cognition need to be made.
7
Introduction
15
fact, the things the novel is about are not present in the situation. A
fictional story and its world are things in the head of the reader (see
Chapter 3). Thus, it seems that a distributed perspective on cognition has little to offer to explain this phenomenon. Also, the thesis
concerning the difference between participatory stories and nonparticipatory stories proposed in this book concerns internal, mental processes—and not behaviour such as physical actions—making
an intra-mental perspective more suitable than a situated/distributed
perspective (the argument is presented in Chapter 4).
Although what is studied in this book is a concrete, real situation, the approach is not applied research. The goal of the book is
not to solve some specific, practical problem, but to obtain general
knowledge about human cognition.
1.3
COGNITIVE SCIENCE AND
INTERDISCIPLINARITY
Cognitive science, the interdisciplinary field of inquiry which
started in the 1950s and received its separate status in the 1970s, has
at its core the task of explaining cognition (Gardner, 1987). The
present book joins this tradition, but not by adopting one key feature of cognitive science which is computational modelling. Rather,
what makes the work in this book fall within cognitive science is,
along the lines of Gardner (1987), its notion of a philosophical focus,
the reliance on mental representations, de-emphasis of emotion, history,
and culture, use of empirical methods, and adherence to interdisciplinarity. The four former features have already been taken up to some
extent and what is left is a discussion of the interdisciplinary aspect
of this book.
The view of cognition as information-processing and the notion
of mental representations rest primarily on the framework of cognitive psychology. The cognitive phenomenon comprehension, as well
as its more specific relative narrative comprehension, has traditionally been studied in cognitive psychology. Cognitive psychology
provides the view that comprehension is cognitive processes operat-
16
Participating in a story
ing on knowledge. In the book, the theoretical framework for participatory stories (Chapters 2–4) builds partly on results from cognitive psychology. Psychological methods were used in the analysis of
memory qualities (Chapter 7), and to some extent the analysis of
spatial mental representations (Chapter 6).
Besides psychology, much inspiration is taken from the area of
linguistics. Linguistic methods and analyses were used in the present
book as a means to get at cognition. Analysis of speech was done in
order to get at people’s spatial mental representations (Chapter 6)
and to study what perspective people adopt on events from participatory stories (Chapter 8). Gesture research provided the framework for the analysis of perspective as revealed through gesture in
Chapter 8.
The nature of fiction has been investigated for centuries in philosophy. Although sometimes highly speculative, philosophical
theories of fiction can provide insight into and serve as a starting
point for empirical studies of comprehension of fiction. In the present book, the theoretical framework (Chapters 2–4) and its constituent concepts are formulated partly using philosophical methods. But philosophy also permeates the empirical part of this book,
in that the questions investigated concern philosophical questions
about mind and reality.
In the area of artificial intelligence, there are implementations of
systems for artificial, automatic comprehension of stories. These
formalisations can give useful information on what a model of natural comprehension might look like. The present book does not
however aim at formalising theory to the extent that it can be implemented computationally. The reason is mainly that too little is
known about cognition of participatory stories, making explorative
empirical studies more suitable.
In literary studies and narratology, the form and functioning of
narrative is studied, mostly using humanistic methods. Literary
studies and narratology provide theories of traditional narratives
that help clarify how participatory stories can be characterised
(Chapters 2–4). The work in present book also draws from earlier
Introduction
17
discussions of the phenomena of participatory stories from literary
studies, film studies, and media studies.
1.4
DELIMITATIONS
It can be very enlightening to state what something is not, so here
follows a discussion of in what ways the work in this book was delimited.
This book is not concerned with how participatory stories can or
should be designed. Thus, it has a scientific rather than an artistic perspective (see Mateas, 1997). Artistic and technological approaches
often focus on the future and on the potential of participatory stories, while this book is concerned with the present forms of participatory stories. The reason for this is straightforward. In order to
empirically study the use of actual participatory stories, they have
to be available for use. The study is an empirical investigation, descriptive rather than normative. Consequently, there will be no elements of literary criticism, where one tries to find grounds for determining the value of cultural products.
This book does not contain discussions of what we learn from
participatory stories, how they can help us in our lives, or what
makes a good participatory story—what is studied is rather the cognition involved when understanding participatory stories and the
outcomes of cognition in terms of mental representations.
Stories and symbols are not discussed from an ontological or semiotic perspective, but rather from a functionalist cognitive perspective.
Note that neither stories themselves (i.e., particular works, such
as books or films) nor participatory stories themselves (such as
computer games) are studied in this book, as is usually done (e.g.,
Buckles, 1985; Aarseth, 1997; Murray, 1997; Ryan, 1997; Wilhelmsson, 2001) (see Chapter 4 for an overview of research perspectives in the study of participatory stories). The study of media itself
tells us little about cognition. In order to study cognition, we must
obtain data about cognition, and that can only be produced by peo-
18
Participating in a story
ple carrying out some cognitive activity, not by studying only the
medium itself. The process of creating participatory stories is not
studied. Rather than the production side, it is the reception side that
is studied.
This book does not deal with issues of computer graphics, computer animation, speech technology, intelligent agents, interface
technology, multi-modal interaction techniques, or other subjects
primarily associated with the technology itself. Rather, in the discussions, it is disregarded to the extent possible whether the
participatory stories are realised in text, images, or sound.
Questions are not investigated concerning biology, neurology,
sociology, politics, or other areas which represent a different level
from the primary study object, namely, cognition.
The emphasis is not on how people use participatory stories, but
on how the processing occurs (Perfetti, 1996). The argument is that it
does not matter for the present investigation whether they are used
for education, entertainment, or something else. The application of
participatory stories for educational purposes is not treated in this
book. Indeed, the general question of the possible influence of
fiction or participatory stories on cognition and behaviour is not
discussed, regardless of whether this influence is considered detrimental (e.g., whether violent computer games increase violent behaviour) or beneficial (e.g., whether children can learn from computer games).
Finally, emotion and aesthetic experience are not considered in
the present study. This is not because they are considered nonexistent or unimportant in connection to the experience of either
non-participatory or participatory stories. The reason is simply that
they do not constitute a necessary part in order to account for the
differences between non-participatory and participatory stories, as
cognition does (see Chapter 4). Emotion and aesthetic experience
are considered analytically separate from cognition and were neither
discussed theoretically nor studied empirically in this book.
Part I
Theoretical framework
2
CHAPTER 2
Stories
in human culture and history. Before
S the advent of ubiquitous
writing systems, all human knowledge was passed
TORIES ARE
on from generation to generation through oral stories, myths, and
re-enactments or rituals and ceremonies. Also in writing cultures,
knowledge and values are represented in stories, still through oral
stories, but also in books, newspapers, film, television, and recently
also through the computer and networks such as the Internet.
But what is a story? There have been theories about stories at
least since 350 B C . In Poetics Aristotle (350 B C /1999) discussed the
art forms of his time. However, he made no attempt to separate the
abstract notion of story from specific art forms that in various way
imitated life. Discussing comedy and tragedy in a mixture of a descriptive and a normative approach, he set out to define what a good
plot is like. There should be unity—a beginning, middle, and an
end—and the events in the plot should follow in probable or necessary sequence. At the beginning of the twentieth century, Propp
(1928/1968) studied structures in Russian folktales, suggesting that
they could be characterised by a limited set of basic functions. It is
not until the middle of the twentieth century that the abstract concept of story (narrative) is described as present in various art forms
and media. The most recent research area that studies stories is narratology. Narratology can be described as the study of the form and
functioning of narrative, or narrative competence (that is, what
knowledge a person must possess to be able to produce and understand narratives) (Prince, 1983). A common definition of narrative
in narratology is that it is the recounting of events by one or more
21
22
Participating in a story
narrators to one or more narratees (Prince, 1988). In narratology narrative is seen as transcending media, so that narrative can be present
in books but also in other media, such as newspapers, television,
and film. Up until the late twentieth century, stories have been seen
mainly as an external phenomenon, existing out in the world, such
as on the pages of a printed book. However, in parallel to the
growth of the cognitive sciences, the notion of story as a mental
phenomenon, existing in the minds of people, has started to spread.
Before continuing, there is a need to address some issues of terminology. The term story will be used throughout this book to refer
to what is sometimes called story and sometimes called narrative.
These terms usually refer to the same thing, but have backgrounds
in different academic traditions. Narrative has been the preferred
term in fields such as literary studies, cultural studies, and narratology. The term story has a background in research in psychology and
artificial intelligence in the 1970s, such as research on scripts by
Schank and Abelson (1977), and on story grammars (Mandler &
Johnson, 1977). It should be pointed out that neither of these terms
makes any assertions concerning the fictional status of events. Thus,
there may be narratives and stories which are fictive and there may
be narratives and stories that are non-fictive. Save a few special,
theoretical meanings of story, there is no reason not to use it instead
of narrative. Story is a simpler word which more easily connects to
general understanding and is therefore preferred. Technical uses of
story will be pointed out as needed.
Another issue is what to call the person or persons who experience a story, if one wants to use a general term for various kinds of
media. Is it the reader, the listener, the viewer? When considering
participatory stories, even more terms come up: Is it the user, the
interactor, the player? The term audience is relatively neutral to
these considerations, as well as to the number of persons who are
involved. Keeping in mind that its passive connotations should be
avoided, the term audience will be used as a general term to refer to
a single person or several persons experiencing traditional as well as
participatory stories.
Stories
2.1
23
A DEFINITION OF STORY AS A MENTAL
ENTITY
The concept of story will here be given a definition in mental terms.
It will be shown how a definition of story as a mental entity can
handle some general problems pertaining to stories better than a
definition as an external entity. A mental definition is also well
suited to act as a building block when defining participatory stories,
as will be seen in Chapter 4. The intention is not to examine how
all the aspects of stories work, but rather to put forward a general,
minimal account of the concept of story.
When theorising about stories, a basic assumption is that much is
gained by being general and capturing what holds for all stories.
The intention in this book is generality—the question of what
makes one story different from another will not be pursued here.
Story is defined as follows:
A story is a mental representation of at least two chronologically
related events, including an actual or intended state change by an agent.
Now let us look more closely at what the definition says.
2.1.1 A mental representation
A story is seen as a mental representation. This representation is constructed in a cognitive system (such as in the mind of a person). The
mental representation can have both internal and external sources.
The mental representation is constructed from any external stimuli
together with cultural conventions of interpretation. The external
stimuli are often constituted of smaller segments of static, external
representations of events. The cultural conventions specify a way
for the external stimuli to be perceived chronologically (‘first this,
then that’), or the stimuli inherently have a chronological order.
For instance, bunches of paper with characters written on them
can serve as stimuli. These characters should be read according to
specific cultural conventions: in Western cultures, left to right, top
to bottom, starting at the top paper and then flipping the pages over
24
Participating in a story
until the bottom one (certain things should be left out, such as publication details and page numbers). This is what we usually call a
book. Note that the symbols on these papers could be read in any
number of ways. There is no inherent chronology or natural linearity in a book. The use of a book is purely conventional. Another
example is small drawings encompassed by squares, scattered across
papers bound together in a bunch (this should be used with conventions similar to the earlier example)—these are comic books. One
example of stimuli where the chronology is inherent is speech. Simply speaking about situations one after the other satisfies the
chronological criteria, but listeners still need cultural conventions to
turn this into a story. Films also have a physically determined chronology, but cultural conventions are necessary for what is shown on
screen to make sense; that is, for it to be experienced as a story. For
the native people of Australia, single paintings on rocks elicit stories, for instance, the telling of a hunting story. The paintings are
interpreted according to cultural-specific rules, creating a story in
the viewer’s mind.
In psychology, complex mental representations called schemas
and scripts have been proposed to account for knowledge in longterm memory. The concept of schema was introduced by Kant
(1787/2004) to account for our general knowledge of the world.
Bartlett (1932) used the schema concept when exploring how narrative comprehension relies on pre-existing background knowledge in
his psychological studies of remembering. When Bartlett’s English
subjects retold an unfamiliar Native American folk tale, it was
transformed according to their own cultural expectations. In the
1970s, scripts were proposed by Schank and Abelson (1977) as a kind
of schemas that incorporate general knowledge about what typically
constitutes common situations, such as the events, roles, and objects
involved in eating at a restaurant. Scripts have been much used in
studies of narrative comprehension, in that they explain how it is
possible to understand both local situations in a narrative and a narrative as a global structure (Mandler & Johnson, 1977). Schemas and
scripts consist of general knowledge, which help us understand the
world by providing expectations for typically occurring items. In
Stories
25
contrast, story, as defined here, is a mental representation of a specific instance. Scripts provide the background knowledge needed
when constructing a story.
2.1.2 Chronologically related events
A story is a mental representation of events. What is an event? Actually, the mental definition of story does not require an answer to
this in an absolute ontological sense. The requirement is that the
cognitive system represents events in some way. Zacks and Tversky
(2001) base a psychological view of events on Quine’s proposal to
regard events ontologically as dynamic objects of bounded regions
of space-time. Psychologically, events would be segments of time at
a certain location that to an observer has a beginning and an end.
For a mental representation to be called a story, it needs to include at least two events. Intuitively, if we are dealing with a single
event, we would call it an event rather than a story. We want a story
to be a sequence of events, which means that there have to be at least
two events. In order to have a sequence, we need to impose another
restriction on the events; that they take place at different points in
time and that the mental representation explicitly states which one
comes before the other.
2.1.3 An agent
We could be content with this first part of the definition: a story is a
mental representation of at least two events related chronologically.
However, upon closer inspection, it allows for event sequences so
simple and uninteresting that we would not want to call them stories. Consider the following case (assume it reflects someone’s mental representation): ‘A leaf blows in the wind. Then a leaf blows in
the wind.’ These are two events following each other. But the
events are not related in any other way, and some further relation
seems intuitively necessary for us to call it a story. It could be argued that a story needs the presence of at least one agent. An agent
could be a person, but it need not be. Bruner (1990) claims that a
story needs to include people, but it seems that an agent could also
26
Participating in a story
be another animal, or even an inanimate object, as long as it is represented mentally as something that can perform actions. For instance, one could imagine a pedagogical story for children, where a
natural force such as wind is given agenthood as it is described shaping the geographical landscape. However, it is not sufficient just to
add an agent in order to have a story. An example building on our
so-called story about blowing leaves above could be: ‘A leaf blows
in the wind. Then a leaf blows in the wind. William is a boy.’ Although this so-called story contains an agent, there seem to be
something missing. An additional requirement is that the agent
must be related to the events.
2.1.4 An intended or actual state change
We may demand that the agent has some intention which involves
one or both of the events. But this cannot be a requirement, because
we would want to accept stories which do not include any intention. Consider this example: ‘Zeki the Zombie lives in a grave during the day. During the nights he climbs out and aimlessly attacks
people walking in the street. But one day he walked off a cliff and
was washed out to sea.’ This seems to be a story, although very
short and somewhat strange. Moreover, it does not include any intentions. If we believe that Zeki is not conscious because zombies
are not conscious,8 and we believe that non-conscious entities do
not have intentions, then the story does not contain intentions (the
same case could be made with stories about robots, as long as we see
them as not having intentions). Thus, intentions cannot be a necessary component of stories. However, if the story told about Zeki
just standing still, doing nothing, we would not call it a proper
story. If intention is missing, instead some sort of state change
caused by actions seems necessary. So, a story could tell about Yuri,
Zeki’s brother, who is alive, pondering about how to rebuild his
house. After the story has told us about these intentions in detail,
we may learn that Yuri decided not to go ahead with his construc8
Zeki is a zombie in the philosophical sense introduced by Kirk (1974).
Stories
27
tion work after all. Thus, no action was performed and no state
change was made. In this case, the relation of the agent to the events
is through the intention. What is considered an intention or a state
change is up to the cognitive system. It need not correspond to conceptions of realistic or even possible phenomena in the world.
2.1.5 Minimal versus typical stories
It should be noted that the definition here gives a bare minimum of
what it takes for something to be a story. Following the definition,
we could end up with a story that is quite far from a typical story. A
typical story would include many more events than two, and also
usually several characters who act according to goals and plans. For
example, the notion of causality—why things happen and why people do what they do—is completely missing from the present definition of story, but is usually incorporated to a high degree in
understanding of typical stories. Of course, the issue of minimal
versus typical accounts is a property of all definitions of real world
phenomena. Thus, the problem of minimal versus typical accounts
of stories is just as prevalent for non-mental definitions of stories.
2.1.6 The disappearance of the narrator
A noteworthy feature of the mental definition of story is that it
does not give any specifics regarding the narrator. As is the case
with Prince’s (1988) definition of narrative above, the narrator has a
central place in narratological definitions of narrative. The narrator
may refer to the author, meaning the physical person who crafted
the story, but can also mean the implied narrator whose voice is
represented in the telling of the story. Does a story always have a
physical person who authored it? Most of the time, but not always.
Consider computer-generated randomly constructed stories (see
Aarseth, 1997; Murray, 1997). In these cases, there simply is no author (in the sense of a person) and it seems meaningless to appoint
28
Participating in a story
an implied narrator if we know that no such thing was put into the
story. 9
Instead, the author and the narrator are seen here as belonging to
a different level from the story itself—the context of listening, reading, or watching a story. In the situation of experiencing a story,
these elements may come into play, so that the construction of the
story by the audience is altered by its assumptions about who the
author and the narrator are.
2.2
DEFINITION OF STORY: MENTAL VERSUS
EXTERNAL
Why not define story as something external, such as what is written
in a book? This surely seems more intuitive and straightforward
than the definition of story as a mental entity presented here. There
are, however, several good reasons to have a mental definition.
First, with any external definition of stories comes the problem
of demarcation. What is a story and what is not? Is the text in a
book a story? One sentence from a book? What about a painting? A
sculpture? Is any physical object a story? No matter how one
defines story externally, one is always begging the question ‘but
what about this, is this a story?’ A definition of story as a mental
representation circumvents this difficulty, since it is not about
physical objects. People can easily construct stories from seemingly
non-narrative stimuli. There may be little to suggest a story from
any physical traces in a medium. A story is created in the response
of external objects only if there are cognitive mechanisms and cultural conventions together with which it can establish a mental representation of events (according to the definition given above), that
is, a story.
It cannot be argued in these cases that the programmer of the computer is the author. Although the programmer set up some general constraints on the form of the
story, she does not know what the contents of the story will be. It seems equally
meaningless to pick the user of the computer as author (when separate from the programmer), just because she initiated the execution of the program.
9
Stories
29
Second, a mental definition of story also allows for stories that
are dreamed up in people’s heads, existing only mentally. Before an
author writes anything down, is it not possible for her to have
thought up the story all in her head? The mental definition also
allows fantasies to be stories. With a definition of story as something external, purely mental stories are not possible.
Third, a mental definition allows for the experience of different
stories for different individuals in response to the same physical
stimuli. In an external definition, there is the problem of establishing what the true story is (since it exists externally, it follows that it
must have some fixed structure). With a mental definition, there is
no need to postulate a single and true story. For instance, people
read the same book but get a slightly different story out of it. The
stories are different because the cultural conventions and each person’s background knowledge vary. Yet the biological and cultural
similarity of people has the result that the story, in most cases, ends
up being roughly the same. If, on the other hand, story is seen as
something external, it becomes difficult to account for individual
interpretations of stories. After all, if there is one and the same story
externally, why does everyone have different understandings of it?
Fourth, the mental definition also explains how different media—
such as printed text, film, and speech—can convey (almost) the same
story. The explanation is that a story can produce roughly the same
mental representations, but through completely different physical
structures. (This issue is further discussed in relation to participatory stories in Chapter 4 in Section 4.2.2.)
Fifth, the mental definition also explains how people can turn
real life events they experienced into stories. With an external
definition, we would need to say that every person’s life story somehow exists out in the world, which seems odd. Rather, the mind
imposes the narrative structure on events in the world, creating a
story. Experiencing stories is a mode of thought, natural to the human mind (in this respect, the work in the present book is in accordance with Bruner, 1990).
One could argue that the mental definition of story is inappropriate because it does not match the general layman’s understanding
30
Participating in a story
of story as an external object. After all, our talk of stories most often implies that stories exist inside books, movies, etc. However,
differences between everyday understandings and scientific definitions abound in science. Consider the physical definitions of time
and space—hardly in accord with people’s everyday understandings.
The whole point of re-defining concepts scientifically is to show
that it is possible to attach a better, more precise meaning to the
term. So, the issue of scientific definitions different from everyday
usage need not concern us, as long as the concepts bear fruit theoretically.
One reason why scholars have stayed away from a mental definition of story may be because it is thought, specifically by scholars
from outside the cognitive sciences, to be difficult to approach
methodologically. The argument goes: An external object is easier
to study than something inside people’s heads, because we can see
and study a book or a movie, but we cannot look inside people’s
heads. However, there are two faults with this objection. First,
‘looking into people’s heads’ is exactly what is the central theme of
cognitive science, the study of what goes on inside the mind, and
there are a variety of successful methods to choose from. This is not
an impediment to a mental definition of stories. Second, external
definitions of story are not secure and objective anyway. Our perception of any external object involves our cognitive processes, and
we have conscious access only to a small number of these processes.
It is vain to believe that one has access to some objective version of
a story just because the stimulus is a physical object such as a book.
Another reason why a purely internal conception of stories may
be resisted is that stories are argued to be social constructions. One
line of argument is to say that stories are created in the act of storytelling. The meaning of the story may even become something
other than what the narrator intended. However, that a story
changes as it is told, perhaps as a result of the social context, is
something that is handled well by the mental definition given here.
For example, an audience listening to a narrator constructs its own
internal story from the physical stimuli in an act of comprehension
which includes background knowledge, partly consisting of social
Stories
31
and cultural factors. Even though the narrator starts with a story as
a mental representation, this representation is itself reshaped by the
act of telling. Selection and elaboration change the original memory, so that the story may become different from what it was, even
for the original narrator.
Finally, an attractive quality of a mental definition of story is
that it is suited for the study of participatory stories. The reason is
that it explains how people can experience stories from seemingly
shattered stimuli, such as the textual fragments of hypertexts or adventure games (further discussed in Chapter 4). In these cases, it is
often difficult to find any coherent physical object that is the story.
Stimuli with little narrative structure can still result in a wellformed story in the perceiver’s mind (see Douglas, 1992, for empirical studies), maybe even fulfilling Aristotle’s requirements of unity.
3
CHAPTER 3
Fiction
A
of stories not discussed in the preceding chapter is
the notion of fictionality. The definition of story as a mental
entity is neutral with respect to fictionality (as are indeed many external definitions). That is, a story could be fictional as well as nonfictional. (Fiction is not used in this book to denote a type of literature, but is used in its philosophical and cognitive sense—which will
be developed below.) To further pin down the concept of participatory stories, the role of fiction needs to be clarified. This is important because people who experience participatory stories treat them
as something that is not real. Participatory stories are fictional.
Fiction is generally held to be a contrast to reality. If something
is fictional, it does not exist in the real world. Philosophers have
been interested in giving ontological accounts of the nature of fiction. The basic philosophical problem is to explain what kind of
things fictional objects are, and how it is possible to talk and think
about them if one claims that they do not exist (e.g., Castañeda,
1989; Van Inwagen, 1977; Parsons, 1980).
Some (post-modern) researchers deny that there is a distinction
between fiction and reality. For instance, Sørensen (1998), in addressing the question of whether computer games are real or
fictional, states that the concept of fiction is a product of modernity
and that it is illusionary. The ontological status of fiction aside, doing away with the distinction between fiction and reality in people’s
understanding of the world is clearly a mistake, because if this was
true, it would lead to considerable behavioural effects in people. If
people thought that a murder actually took place while watching a
N ASPECT
33
34
Participating in a story
thriller movie, or that they themselves actually killed someone
while playing a computer game, they would indeed behave very
differently from what can now be observed. The distinction between fiction and reality is a crucial element in how people comprehend the world.
The notion of fiction used in this book tries to avoid the philosophical issues and instead focuses on this functional role of fiction.
3.1
A DEFINITION OF FICTION AS COGNITION
Let us look at the definition of fiction for the purposes of this book:
The content of a mental representation R is fiction to a cognitive
system C if and only if C believes that R should not be evaluated in relation to the real world.
A cognitive system is usually a person (but the definition is open for
other entities, such as computer systems). That something should
not be evaluated here means slightly different things depending on
the content of the mental representation. If it is an object or an event
(an event can be considered a type of object, as discussed in Section
2.1.2), the cognitive system attributes non-existence in the real
world, that is, the object does not exist or the event did not occur. If
it is a proposition, such as ‘The tooth fairy works at a bank’, the
cognitive system does not attempt to determine the truth value in
relation to the real world.10
Thus, fiction here does not have the usual ontological meaning. In
other words, no attempt is made to define what the nature of fiction
is. Instead, a relativistic notion of fiction is taken, establishing it as a
binary predicate involving a cognitive agent and some mental content. This definition of fiction is similar to that of Rapaport and
Shapiro (1995) who call it an ontological epistemological notion of
Although this example also has several interpretations which have truth value in
the real world, the intended interpretation is that it is about an imaginary character.
10
Fiction
35
fiction to point out that what is concerned is a cognitive agent’s understanding of fiction, not fiction as an absolute ontological concept.
The same view is adopted by Chafe (1994), studying people’s telling
of stories. Making this move avoids much of the difficult ontological discussion about fictionality.
Fictionality is defined as a mode of thought. Something can be
considered fictional or not, and the claim is that people know how
to do this spontaneously. What is considered fictional or not is determined by pragmatic rules of usage. For instance, the covers of
fiction books look systematically different from non-fiction books.
Documentaries present content differently from drama. There are
also linguistic clues to what should be interpreted as fiction, such as
‘Once upon a time…’ or ‘Let me tell you a story…’ in oral storytelling. Clues such as these and many more aid people to interpret
something as fiction or non-fiction. Of course, the division can be
blurred so that it is hard to determine if something is fiction or not.
And of course, people do sometimes make mistakes, for example, so
that they interpret something as non-fiction when it was intended
to be fiction (e.g., the 1938 radio broadcast of War of the worlds
which made people believe that Martians actually were invading
Earth (Bartholomew, 1998)). Still, this is the exception rather than
the usual case. The context usually leads people to the intended
interpretation. Regardless of occasional misjudgements, there are
mechanisms in the human cognitive system which work to keep
reality and fiction apart.
Now it is possible again to raise ontological questions. But this
time, the questions are about the ability to experience fiction, rather
than questions about fiction itself. The issue can be approached
from several viewpoints. From a Darwinist evolutionary perspective, one could ask: How did the ability to experience fiction
evolve? Why did it evolve? When in the prehistory of humans did it
evolve? Are humans the only animal with this ability? And from a
developmental perspective, one could ask when the ability to separate fiction from reality develops in the child. Considering a neurological perspective, one could ask: Are there special neural mechanisms or brain regions which maintain the separation of fiction
36
Participating in a story
from reality? From a pathological perspective, are there disorders
where we lose this ability (schizophrenia appears to be one)? Even
though these are highly interesting questions to ask, they will not
be addressed in this book. Instead, the ability to experience fiction
will be viewed from a functional perspective, which is more relevant with regard to the questions asked in this book. What does it
do? How does it affect behaviour? How is the phenomenon of narrative understanding affected by the fact that the understander treats
some information as fictional? What would happen if the distinction between fiction and non-fiction were not made?
3.2
SOME OBSERVATIONS ON THE
COGNITION OF FICTION
A number of observations regarding the cognitive processing of
fiction will now be introduced. These observations will subsequently be used as points to consider when discussing theories of
the cognition of fiction. It is argued that any successful theory of
cognition of fiction should be able to explain these observations.
The first observation is that there is a transfer of background
knowledge to the understanding of fiction. The second observation
concerns the opposite: transfer of fictional information to background knowledge. The third observation is that there are not necessarily any perceptual or sensorimotor differences between fictional and real events and actions. Finally, the fourth observation is
that the distinction between fiction and reality can sometimes be
weakened in memory.
3.2.1 Transfer of background knowledge to
understanding of fiction
In the understanding of fiction, people need to use existing background knowledge, including knowledge about cultural, social, and
physical matters. For instance, a unicorn can be imagined given
knowledge about horses and horns. However, it is not only exotic
Fiction
37
cases of fiction where the audience needs to provide background
knowledge. It comes into every part of understanding fiction. Basic
things, such as that ‘Sherlock Holmes is a human’, demand general
knowledge about what it means to be human. However, the problem is not to use any background knowledge, because then we
would hastily conclude that the statement ‘Sherlock Holmes is a
human’ is false, since there is no such thing as Sherlock Holmes in
the real world (Rapaport & Shapiro, 1995).
3.2.2 Transfer of fictional information to background
knowledge
A second observation is that the influence of information also may
be in the direction from fictional information to general background
knowledge about the world. For instance, clearly, things about medieval France can be learnt by reading a novel about medieval
France, even though the novel is purely fictional. Real-world beliefs
can be influenced by exposure to fiction (Prentice & Gerrig, 1999).
However, as discussed above, the distinction between fiction and
reality is still upheld at a general, conscious level. They are not
treated equally. The question then is: When and how does fictional
information enter background knowledge?
3.2.3 The absence of perceptual differences
A third observation concerns the conditions of making an attribution of something as fictional or real. The observation is that there
need not be any perceptual or sensorimotorical differences between
fictional and real events and actions.
Let us consider how fiction functions when an audience sees a
play on stage. When someone is murdered in the story, the audience
should recognise it as a murder. If they did not, they would not be
able to understand what was going on in the story. But they should
not believe that a murder really happened and rush out and call the
police. Note that the physical stimuli, that is, the action on stage,
might be perceptually indistinguishable from the real thing (a glis-
38
Participating in a story
tening knife, fake blood, etc.). (What matters here is what is perceived, not what happens in any objective sense.) Fiction plays an
important role, not only in perception, but in action as well. Let us
consider an example of a video game arcade where there is a video
game incorporating a gun. The game is played by firing the gun at
characters on a video screen. The gun looks and feels very much
like a real gun, and is handled in the same way. Now, not only is
the performed action on a conceptual level identical to the real
world (‘shooting someone with a gun’), but the action is also sensorimotorically identical. The sensorimotor activity of holding the
fake gun is identical to holding a real gun. The only thing that keeps
the player from believing that she actually is shooting real people is
a belief that it is not real, but fictional. One could give other examples with even more sensorimotor and physical similarities to the
real world, such as when children are engaging in pretence play, or
when participants are involved in a live role-playing game. In the
same way as in the example with the murder on stage, these are
cases where no perceptually noticeable physical differences exist11.
3.2.4 Blurring the distinction between fiction and nonfiction in memory
People are generally good at remembering at a later stage which
events and actions were fictional and which ones were non-fictional.
For instance, in the play example they remember that the murder
was fictional but that, say, the announcement of a pause was real. In
the video gun example, they remember that they did not actually
kill anybody, but they remember if they earned a position on the
list of high scores.
Sometimes, however, the distinction between fiction and nonfiction can be obscured, especially as time passes between the event
Whether a situation presents something fictive or real often leads to different objective consequences—such as that you are arrested for a real murder but not for a
fictive one—but consequences cannot play a key role since they are not known at the
time when the events are experienced.
11
Fiction
39
and when it is later remembered. People may in some cases mistakenly believe that a fictional event was real. The distinction between
fictional and real events is fallible.
3.3
THE COGNITION OF FICTIONALITY
It follows from the above discussion that there must be some difference represented in memory that allow people to separate fiction
from reality. How is fictional information represented in the cognitive system? As the final part of this chapter, this question will be
addressed by first looking at how the distinction may be approached
conceptually. Then, candidates for a theory of the cognition of
fiction will be discussed.
3.3.1 Possibilities for cognition of fictionality
There are several logically possible explanations for how people are
able to remember which events were fictional and which events
were non-fictional. Consider the following alternatives:
(i)
It could be that the contents of the events themselves reveal
whether they are fictional or not (e.g., fictional content is unrealistic or otherwise unique), such as unicorns, monsters, and
mountains of gold.
(ii)
With the mental representation of each event stored in memory there could be some kind of marker which flags the event
as fictional or real.
(iii) A similar position to the one above is that there may be an
association from the memory of the event to its source, for instance the memory of a certain car chase could have a link to a
memory of a certain action movie.
(iv)
The events are mentally represented in memory in conjunction with the context (e.g., time, space, sensorial) in which
they occurred, which lets a person deduce at the time of recall
40
Participating in a story
whether the event occurred in a typical fictional context (as
when sitting in a theatre or while reading a book). This position would allow the highest number of errors separating reality from fiction, since there may sometimes be limited contextual information available, which could lead to a misjudgement.
(v)
A combination of some or all of the above.
These possible explanations will be used in the following overview
of theories of the cognition of fiction, and some associated cognitive
theories. In surveying the literature, it is clear that the topic of
fiction has received very little attention within cognitive psychology
and cognitive science in general. For instance, all empirical studies
that were found were dated 1989 or later, even though many other
aspects of cognition have been studied extensively since the 1950s.
The reason for this situation can only be speculated about. Possibly,
cognition which concerns the organism’s relation to the real world
(such as perceiving, remembering, and reasoning about the world)
has had precedence in research. Perhaps fiction has been considered
to be merely recreational, play, something not useful. Perhaps the
topic has been avoided because of its philosophical difficulties.
The following discussion of cognitive theories will involve reality
monitoring, compartmentalisation, pretence, and computational
modelling of the understanding of fiction.
3.3.2 Reality monitoring
A theory which addresses a phenomenon similar to fiction is reality
monitoring (Johnson & Raye, 1981). The theory explains how the
cognitive system works when separating memories of events that
actually happened from events that were merely imagined. According to the theory, there are several types of processes at work, at
both an unconscious and a conscious level. In part, the content of
the memory influences the decision (which corresponds to position
1 in the list above). Mainly, there are processes which evaluate the
context, in terms of the presence of spatial, temporal, perceptual,
Fiction
41
emotional, and other details in the memory. Memories with a
higher degree of contextual details are more likely to be attributed
to a real event, and memories with a higher degree of mental operations, for example, thinking and reasoning, are more likely to be
classified as an imagined event.
The theory does not address the question of fiction directly, but
if a parallel is made between fiction and imagined events, it becomes
a theory about how people can separate memories of fiction from
memories of reality. It can be seen that the reality monitoring theory adheres to position 4 in the list presented of possible explanations of the cognition of fiction. Returning to the four observations
discussed, the reality monitoring theory is well equipped to handle
Observations Two and Four, since the reality monitoring processes
work heuristically; some memories may be misclassified as real
when they were actually fictional. The theory is neutral in relation
to Observation One. However, the theory has problems accounting
for Observation Three (the absence of perceptual differences). For a
priori reasons, such as the video game gun and the live-role playing
examples above, the theory does not seem to capture the cognitive
aspects of fictionality. In general, the temporal, spatial, and perceptual context need not differ between the fictional and the real cases,
which is central in the theory of how actual and imagined events are
separated in memory. The application of reality monitoring theory
to cognition of participatory stories is empirically explored in
Chapter 7, where qualities of memories from participatory stories
are investigated.
3.3.3 Experimental studies of compartmentalisation
Potts, St. John, and Kirson (1989) specifically studied representation
of fictional information. They used Anderson’s ACT* semantic
network theory (although they stress that their discussion holds for
any memory model), which in part consists of concepts nodes with
associative links between nodes. According to Potts et al., fictional
information is stored through compartmentalisation, which means
that it is stored separate from, not linked to, general world knowl-
42
Participating in a story
edge and instead has links to the fictional story source. Thus, the
position proposed by Potts et al. corresponds to number 3 in the list
of possible explanations above. Interpreting the theory literally, it
cannot account for Observations One, Two, or Four. It is neutral
with respect to Observation Three.
Marsh, Meade, and Roediger (2003), building on the work of
Potts et al., investigated three hypotheses about the representation
of fictional information using a series of experiments: (i) Fictional
information is completely integrated into the person’s world
knowledge, or (ii) it is completely compartmentalised, or (iii) fictional information is represented according to a hybrid model,
where fictional information is associated with both world knowledge and its fictional source. Marsh et al. also subscribe to position 3
in the list of possible explanations above. Evidence was found for
the hybrid position. Marsh et al.’s theory allows all four observations (it is neutral with respect to Observation Three), but does not
explain any of them.
There are some features of the method used that makes generalisation to authentic contexts difficult. Marsh et al., like Potts et al.,
used artificial texts and did not study reading for pleasure. Because
of the experimental design, the participants most probably anticipated subsequent memory tests, which would affect how they
monitor their reading and how much time is spent on rehearsal.
This could potentially make the situation very different from a
situation where they read for pleasure. Generalising to more authentic reading situations should be done very carefully. Another
remaining problem is that despite clear experimental results, it is
not known why associations are sometimes formed more strongly to
world knowledge and sometimes more strongly to the fictional
source. Also, very little is known about details of the actual representations (Marsh et al., 2003).
3.3.4 A cognitive theory of pretence
On a slightly different but related issue, Nichols and Stich (2000)
proposed a theory of how beliefs are represented mentally when a
Fiction
43
person engages in pretence. Pretence means acting as if something
exists (which does not exist), or that an object is something other
than what is actually is, such as that a banana is a telephone. Although pretence and fiction may appear to be two different things,
they are closely related. A fiction is something that does not exist in
the real world, but in the mind, such as ‘this banana is a telephone’,
or ‘there exists a world in which I am king’. This is exactly what
pretence is. Nichols and Stich’s theory is presented at a rather
coarse level of detail and they simply assume that, in the cognitive
system, there is a ‘pretence box’, just as there are ‘belief’ and ‘desire’
boxes, in which propositions can be put. Thus, they hold position 2
from the list above. A problem in this approach is that if the cognitive system generates inferences from beliefs in both the pretence
box and the belief box, contradictions and false conclusions easily
arise. For instance, although we pretend that a banana is a telephone, we should not make the inference that the banana could be
used to call an ambulance in a real emergency. The possible contradictions that could arise when inferences are drawn from both the
‘real world’ and the ‘pretence world’ (such as ‘this banana is a telephone’) are handled in the theory by a filtering mechanism, which
is not further specified. Nichols and Stich (2000) confess that very
little is known about how inferences are filtered, but they note that
there is empirical evidence that it works. Thus, Observations One,
Two, and Four are acknowledged, but not explained, by the theory.
It is neutral regarding Observation Three.
3.3.5 A computational model of understanding of
fiction
There has been at least one preliminary proposal that goes into the
details of representations and inferences of fictional information.
Rapaport and Shapiro (1995) implemented a computational model
of a story-understanding agent, Cassie, based on the S N e P S semantic network belief representational and reasoning system. The agent
read a fictional story and could use world knowledge in understanding the fictional content, and the fictional content could in some
44
Participating in a story
cases become integrated with world knowledge. As a fictional story
is read, Cassie sets up a story space by linking the contents of the
story to a story operator. The story space is formally equivalent to
Cassie’s other belief spaces, such as representing ‘Mary believes that
John is a boy’. Two rules are used to control migration of facts and
inferences inside and outside the story context. Related to Observation One is that propositions from outside the story context are
assumed to hold when necessary for understanding the story and
could be withdrawn at a later stage according to Cassie’s belief revision system. Similarly, related to Observation Two, is that propositions from within the story are assumed to hold outside the story
context when necessary for understanding the real world, and these
propositions can in the same way be withdrawn at a later stage. One
way of seeing it is that Rapaport and Shapiro (1995) add detail to
what Nichols and Stich (2000) call the filtering mechanism. However, Rapaport and Shapiro note that their work is preliminary and
needs further implementation. The theory is consistent with Observation Three. Although the theory does not address Observation
Four, it does provide a mechanism which allows the distinction
between fictional and real events to change.
In summary, the cognitive process of comprehending fiction is not
well understood. Out of research on reality monitoring, pretence,
and representation of fictional information in memory, the most
promising attempt appears to be the computational approach of
Rapaport and Shapiro (1995) which represents a story in a story
context by linking the story facts to a story operator and gives some
suggestions as to how information can both come from world background knowledge into story understanding, and transfer from stories to world knowledge. However, as can be seen from their approach, people’s performance of handling fiction presents a complex
computational problem.12
Advances in the understanding of the cognition of fiction may take time—neither
Nichols and Stich (personal communication, April 6, 2000) nor Rapaport and
12
Fiction
45
Giving an account of fiction is important because people treat
participatory stories as fictive. Instead of providing an ontological
characterisation of fiction, the concept was approached here as
something considered by a cognitive system as not relating to the
real world. Distinguishing fiction from reality is by no means an
immaculate process, but it works most of the time, as people’s behaviour shows. Compared to traditional stories, fiction takes on
additional importance when participatory stories are considered,
because these not only represent but also invite to action in a fictional world. With cognitive notions of stories and fiction, the
foundation has been laid for a cognitive approach to participatory
stories—the focus of the next chapter.
Shapiro (personal communication, February 19, 2003) plan to continue their research
on these issues.
4
CHAPTER 4
Participatory stories
S ing them are not a new phenomenon. A parent telling a child an
TORIES THAT
can in some way be altered by the one experienc-
invented bedtime story might receive input from the child about
what is to happen as the story unfolds. Another example is that
even very young children spontaneously engage in pretence play, by
themselves and with other children. Since the 1970s, people have
enjoyed role-playing board games (e.g., Dungeons and dragons, and
many others that followed in its path). In a board role-playing game
a story unfolds as a result of a so-called dungeon master, a person
who keeps track of the story, and the players, who portray characters in the story. There are also other role-playing games, called live
role-playing, where the participants act out and influence a story
together, in real surroundings, such as a forest, often during extended periods of time such as for several days. These are all examples of stories being influenced by the ones experiencing them,
which do not presuppose advanced technology. It is more recently,
however, that participatory stories have spread to a wider audience,
namely, with the introduction of the personal computer. The computer is at the heart of what is sometimes called interactive media.
In the early 1980s, a form of computerised purely textual interactive
narrative fiction known as adventure games was a commercial success. Adventure games build on the non-computerised role-playing
games, with the difference that they involve a single player. These
were eventually replaced on the market by graphical adventure
47
48
Participating in a story
games in the late 1980s.13 As graphics and audio technology developed, multimedia programs that combined stories and sound, image, and text became available. A way of organising information
called hypertext was also promoted by the spread of the personal
computer. Originally a way to structure information by tying together pieces of information with associative links, hypertext can be
used to tell stories where the audience controls the order in which
textual fragments are read. The accessibility of digital sources increases rapidly as the Internet continues to grow. On the Internet,
stories are created by the actions of the participants in M U D s (multiuser dungeons), which are text-based multi-player fictive environments, and their graphical counterparts. There are also various applications of virtual reality that incorporate stories, both on the
Internet and elsewhere. Participatory stories have also been used for
various applications, for example, in education and as an evaluative
tool in job application situations to obtain psychological profiles of
the applicants. There are attempts to merge the television with the
personal computer to produce a new medium for entertainment for
the general public, which then holds the possibility for stories that
the audience is able to influence.
This brief historical overview has introduced examples of stories
which allow some kind of influence on the part of the audience. In
this chapter, the concept of participatory stories will be considered
in some detail. The two preceding chapters gave definitions of stories and fiction as cognitive phenomena. This is the foundation on
which the concept of participatory stories is built. First comes an
overview of how the phenomenon of participatory stories has been
studied by researchers in various fields. Next, four fundamental
questions for a theory of participatory stories will be addressed—a
discussion that will conclude with a full definition of the concept of
participatory stories: (i) the question of demarcation of participa-
However, the creation and playing of textual games—referred to as interactive
fiction—have not stopped with the advance of graphical adventure games, but have
been and still are being created continuously. See, e.g., the Interactive fiction archive
at http://www.ifarchive.org/.
13
Participatory stories
49
tory stories from other kinds of stories; (ii) the role of fidelity and
multimodality in participatory stories; (iii) the question of what it
means to participate in a story; and (iv) in what sense participatory
stories are fictional.
A note on terminology is in order. Since the term participatory
stories is introduced here, it is not used by the other approaches discussed. Other approaches use other terms and often even use
slightly different underlying concepts when discussing stories that
can be influenced by the audience. Nevertheless, the term participatory stories will be used to talk about how various approaches handle
the phenomenon.
4.1
THE STUDY OF PARTICIPATORY STORIES
Starting in the 1980s, researchers from many different areas have
investigated stories in interactive media, from very different viewpoints. Few other phenomena display such a variety in how they
are studied regarding goals, objects of study, and methods. Just
glancing at the research literature, it may not be obvious that this is
the case. It is most often the case that researchers do not state their
goals, objects of study, and methods explicitly, which might lead to
a view of the collective efforts as falsely homogeneous. For instance,
the goal, the object of study, and the methods are all different in a
cognitive perspective compared to a technological perspective, although both use ‘interactive fiction’ or ‘interactive narrative’ as
keywords when describing their research.
One possible way of grouping earlier approaches will now be
presented, divided into technological, artistic, literary theoretical, social, psychological, pedagogical, and cognitive perspectives. The purpose is to present the different perspectives held and to show how
they are different, rather than to give an exhaustive survey of the
area. Some researchers are mentioned to serve as examples of the
perspectives. Often, researchers combine more than one perspective
50
Participating in a story
Aarseth
Hypertext theory
(Landow, Bolter, et al.)
Douglas
Ryan
Holmquist
Literary theoretical
Technological
Cognitive
Pedagogical
Oz Project
Davenport
Plowman
Artistic
Murray
Psychological
Social
Figure 4.1. Perspectives on stories in interactive media. A possible grouping of approaches showing the most prominent overlaps. Selected researchers associated with
the perspectives are also shown.
in their research. The purpose of this exercise is to specify the location of the approach taken in the present book among other approaches that have been made, and to see in what ways it is different. The various perspectives not adopted in this book will be presented briefly, followed by a more in-depth look at how a cognitive
perspective differs and what has been done using this perspective. In
Figure 4.1, the perspectives are presented graphically as sets, with
some researchers working within them.
4.1.1 Research perspectives
Within a technological perspective, people design and develop technology, systems, architectures, and computer programs for stories
in interactive media. The goal is to implement actual systems as well
as to find general principles (e.g., Holmquist, 1997). Disciplines involved are mainly computer science, systems science, and engineering. In interaction design, a technological perspective is combined
Participatory stories
51
with an artistic perspective. Researchers within the area of humancomputer interaction are concerned with adapting the systems to
human thinking, so that they become easy to use (e.g., Shneiderman, Kreitzberg, & Berk, 1991). Even if areas such as psychology
are involved in such studies, the goal is still to develop systems or
principles for developing systems, and therefore it is a technological
perspective. The technological perspective is one of the dominant
perspectives on stories in interactive media, probably because the
computer is the medium, and computers have traditionally been
studied within these disciplines.
Stories in interactive media can be investigated or produced as an
art form, with the goal being either to learn their possibilities for
artistic expression, or to produce an aesthetic experience in itself.
This artistic perspective is often combined with a technological perspective. When it comes to theorising about using the computer to
support such stories, an early approach is Laurel’s (1991) book
Computers as theater, in which she sketches an ‘Interactive fantasy
system’ where the drama is affected in interaction with a user. The
Oz Project at Carnegie Mellon University (Mateas, 1997) generated
both theories and actual systems, mostly text-based, about what
they call ‘interactive fiction’. Other approaches in an artistic and
technological perspective are the Interactive Cinema Group at MIT
(Davenport & Murtaugh, 1997), and Janet Murray’s (1997) research,
presented in the book Hamlet on the holodeck: the future of narrative
in cyberspace.
In a literary theoretical perspective, theories of literature (and,
sometimes, cinema) are applied to narratives in interactive media,
where questions are asked such as: Are they literature, and if so,
what kind of literature? The main goal in the literary theoretical
perspective is to give a systematic account of these narratives (e.g.,
Espen Aarseth’s, 1997, book Cybertext). Within hypertext theory,
literary theory is combined with artistic, technological, and (often
speculative) psychological perspectives, often with political goals
(e.g., Landow, 1997), or discussion of new aesthetic values
(Moulthrop, 1991). Ryan (1997) combines literary theory with a
technological perspective while discussing possible architectures for
52
Participating in a story
stories in interactive media. The adventure game was studied early
from a literary perspective by Buckles (1985) and Sloane (1991).
The relationship between stories in interactive media and society
is the focus in a social perspective. Topics considered are, for instance, effects on politics (e.g., Landow, 1997), gender issues (e.g.,
Cassell & Jenkins, 1998), identity (e.g., Turkle, 1995), and mass media consumption (e.g., Whitaker, 2001). The disciplines involved are
mainly sociology and cultural studies. The goal is to account for,
and often to provoke change of, social structures in society.
From a psychological perspective, stories in interactive media can
be studied concerning, for instance, what emotional effects they
have and how their rhetoric works. The goal is to formulate general
theories of reactions to narratives in interactive media. Two subordinate perspectives within a psychological perspective where research has been carried out are pedagogical and cognitive perspectives.
A large study area is the educational use of interactive media, such
as hypertext and interactive multimedia. The goal in a pedagogical
perspective is to ascertain whether and how these new narratives
improve learning, and to produce systems or guidelines for developing systems for learning (e.g., Plowman, 1996a).
After this sketchy overview of perspectives, the cognitive perspective will be considered in more detail.
4.1.2 A cognitive perspective
Within a cognitive perspective, it is the thinking that is studied. The
disciplines involved are cognitive science (see Gardner, 1987, and
also the discussion in Chapter 1) and discourse psychology (see
Graesser, Swamer, & Hu, 1997 for discourse psychology of traditional stories). The goal is to give a general account of the thinking
processes used in comprehension of stories in interactive media, for
example, what kind of inferences are drawn by persons during
comprehension, or how mental models may differ from traditional
stories (e.g., Foltz, 1996). (Note that the cognitive perspective is not
a usability perspective; the latter has as a goal to produce or improve
Participatory stories
53
systems, while the former aims at formulating general theories of
human cognition.)
4.1.2.1 A cognitive perspective compared to other perspectives
Let us first look at why a cognitive perspective presents a different
approach from the other perspectives. A cognitive perspective differs from other perspectives in several ways, regarding research goal,
study object, and methods. The goal differs from technological, artistic, and pedagogical perspectives in that there is an attempt to
model a natural phenomenon, namely, comprehension. A cognitive
perspective is not a technological perspective. Both perspectives rely
on the understanding of the nature of stories in interactive media,
but there the similarities end. In a technological perspective, one’s
models are not models of anything—they are simply built to be applied (to solve some problem or to entertain an audience). In a cognitive perspective, on the other hand, one’s models are models of
the comprehension as a natural phenomenon. Not only the goals,
but also the object of study and methods are different in these two
perspectives. The literary theoretical as well as the social perspective
are both different from a cognitive perspective in that they do not
share its object of study. The primary object of literary theory is
literature or the text, while sociology studies the behaviour of
groups. In contrast, the study object in a cognitive perspective is the
process of comprehension on an individual level, that is, the thinking. In a cognitive perspective, there are no attempts to develop systems. Just as cognitive psychologists studying text comprehension
are not novelists, research into interactive story comprehension is
not about authoring participatory stories.
Concerning method, the cognitive perspective differs from most
other perspectives in that it is about theory and model building,
which are evaluated empirically, typically using experimental methods. Finally it is concerned mostly with basic science, in contrast
with technological and pedagogical perspectives, which typically are
applied sciences. It can be argued that the cognitive perspective
sketched here is better than most other perspectives for methodological reasons, because it satisfies general criteria of a scientific ap-
54
Participating in a story
proach better than the other approaches discussed. Describing the
phenomenon of participatory stories is something a cognitive perspective shares with most other perspectives, but by including
causal mechanisms, a cognitive model can provide explanation for
why things happen. This is a feature lacking in most other perspectives, such as exploring the artistic potential of participatory stories,
or collecting rules of thumb for how to build computer systems
that support participatory stories. By explicitly stating elements of a
model, it can be checked so that it does not contain contradictions,
and that it is reasonably complete. Models in other perspectives, if
they are formulated at all, rarely share these features. Unlike many
other perspectives, cognitive theory involves the principle of
falsifiability and can be evaluated by checking it for correspondence
with empirical findings.
4.1.2.2 Earlier research in a cognitive perspective
There have been very few studies of comprehension of stories in
interactive media. Douglas (1992) studied how lack of coherence in
stories affects the audience. Her main results are that people strive
for and see coherence even when someone has tried to remove all
coherence in a narrative (for instance, by cutting out and randomly
assembling all sentences in a story text). Although the study concerned printed stories, she particularly discusses the implication for
hypertext stories. (This is another support for the mental definition
of story given in Chapter 2.) Plowman (1998) studied cognition and
learning in educational stories in interactive media. Her work aims
at producing design guidelines for development of educational multimedia (such as that video sequences are a way of increasing narrative coherence; Plowman, 1996a), and accounting for what she calls
multimedia literacy (which mostly concerns issues of usability of the
user interface; Plowman, 1996b). She does not give a general explanation of comprehension of narrative in interactive media. Risden
(1997) applied a theory from the study of traditional story understanding, called Causal Network Theory, in a small-scale explorative study using a hypertext story. The study, although too small
Participatory stories
55
to provide a basis for any real conclusions, indicated that people use
similar causal reasoning in hypertext stories as in traditional stories.
On the other hand, in a cognitive perspective on traditional stories, comprehension has been extensively investigated in the psychological study of the reading process. There is a lot of knowledge
about what happens at a detailed level when a person experiences a
traditional story (e.g., Kintsch, 1988). Why are general theories of
comprehension of stories in interactive media almost non-existent?
First, it could be that stories in interactive media simply have been
overlooked—researchers have not been aware of their existence.
Second, stories in interactive media could have been considered uninteresting or not worthy of serious study. Third, the methodological task of empirically comparing them with traditional stories
could have been judged too difficult. Fourth, stories in interactive
media might have been neglected because they are considered to fall
within theories of traditional stories. In other words, stories in interactive media might have been viewed as just another instance of
traditional stories, and therefore not in need of a separate explanation.
Let us see how these four points can be countered. The first and
second reason can be argued against by noting that stories in interactive media exist and are widespread, especially, as is argued in this
book, that they encompass a larger class of phenomena than considered by earlier approaches, and not only ‘electronic’ or ‘digital’ stories. Further, technological development is likely to spread participatory stories even more in the future. The methodological difficulties mentioned as the third point above are definitely real, but they
constitute no real reason not to begin investigation. Even though it
may not be possible to set up experimental studies with the
sufficient level of control necessary to draw conclusions based on
quantitative analysis, other methods can be used. Explorative
methods of both on-line (while comprehension takes place) and offline (after comprehension has taken place) aspects can be used. Online aspects can be investigated by studying behaviour (observations
or analysis of logged interactions) or speech (using think-aloud
methods) when participating in a story. Off-line aspects can be stud-
56
Participating in a story
ied by exploring people’s long-term memory (e.g., via language)
from a variety of viewpoints, such as perspective in episodic memory and spatial mental models. The fourth reason against studying
comprehension of participatory stories—that they are just the same
as traditional stories—is countered in this book from both a theoretical and an empirical viewpoint. The section below on what it
means to participate in a story shows how traditional stories differ
conceptually from participatory stories. Empirical findings that
challenge the fourth reason are presented throughout the rest of the
chapters in the book.
This overview of perspectives shows that participatory stories
have been studied from many different perspectives, differing in
goals, methods, and object of study. The cognitive perspective of
participatory stories is well suited for the goals of the present book,
and links naturally to the mental definitions of story and fiction
from the preceding chapters. Only limited work has been done in a
cognitive perspective, so that is an additional factor that motivates
the present work. In the next section, the cognitive perspective will
be further fleshed out and an important point will be made: When it
comes to describing what makes some stories participatory and others not, it is necessary to adopt a cognitive perspective.
4.2
SOME FUNDAMENTAL QUESTIONS FOR A
THEORY OF PARTICIPATORY STORIES
Any theory of stories that can be influenced by the audience (participatory stories, or whatever similar term is used) would need to
address at least the following basic questions:
(i)
How is the demarcation made between participatory stories
and other stories? What criteria are used? What instances (i.e.,
individual works: books, films, computer programs, etc.) are
considered participatory stories and which are not?
(ii)
What is the role of fidelity (accuracy of the representation)
and multimodality (combining several sense modalities, such
Participatory stories
57
as vision, hearing, and touch) in participatory stories, for user
input or for system output?
(iii) What does it mean to participate in a story? In terms of interaction, exactly what is it the audience interacts with?
(iv)
Are participatory stories fictional? If so, in what way?
Each of these themes will now be discussed and it will be shown
how some other researchers have approached them. The questions
will be used to structure the discussion of other theories and work
carried out concerning participatory stories. The discussion will
result in a definition of participatory stories as used in the present
book. Along the way, the key role of cognition will be evident in
the answers provided to all these four questions.
4.2.1 Demarcation—drawing a line between
participatory stories and other phenomena
4.2.1.1 To demarcate or not to demarcate
When approaching a number of instances, such as individual works
of novels, short stories, movies, and computer games, there are two
fundamental ways of doing the description. One way is to set up
criteria for sorting the instances into categories based on some similarities and differences, such as the main concepts participatory and
non-participatory stories, as is done in this book. The other way to
approach the task is to claim that there really are no differences between instances, and that they all share the same qualities.
Concerning participatory stories, the latter way is adopted by
Alexander (1999). In his book Screen play, he discusses how new as
well as old media are constituted by play in the interpretative act of
the audience. He does not make a distinction between participatory
stories and non-participatory stories. Interacting with a computer
game involves play. Watching a movie involves play. Listening to a
story involves play. However, in stretching the concept of play to
its limit to include all forms of media, there is a risk of losing its
58
Participating in a story
meaning and usefulness. Also, Alexander misses one of the philosophically and cognitively most interesting things about play: how the
organism can distinguish between play and the ‘real world’. Another way to deny that there is a difference between participatory
and non-participatory stories is to claim that all texts are interactive
because meaning is created in an interactive process with the reader.
Agreed, but in addition to the interpretative act, which is present
for all texts (and for perception of all kinds, for that matter), participatory stories add another interactive process (more on this in
section 4.2.3 below).
Most other approaches to participatory stories assume that there
is a difference from traditional stories, as will be discussed. The
question then is what criteria are used to draw the line of demarcation. Another issue concerns granularity. How fine a division does
one want to make? How many categories does one use to sort instances into? If we use too few categories, we may miss ways in
which instances differ. If we use too many categories, the classification does not provide enough overview, which would run counter
to what we set out to do in the first place. In the following
discussion it can be seen that researchers have chosen various sizes
of granularity. A discussion now follows of the demarcations of
participatory stories by Murray, Ryan, and Aarseth, in that order,
and end with the demarcation introduced in this book.
4.2.1.2 Murray’s demarcation
Janet Murray’s (1997) book Hamlet on the holodeck: the future of narrative in cyberspace can be described as a hopeful vision of the future
of various forms of what Murray calls ‘cyberdrama’ (and sometimes
‘electronic narrative’). Murray points out that reading and viewing
traditional narratives are not passive activities (she finds support in
the reader response critics Holland, Iser, and Eco), so the distinguishing criteria could not be based on passive versus active. Nevertheless, she argues for a difference between traditional narrative and
‘electronic narrative’, although the argument is very brief and
incomplete. All she says about the matter is:
Participatory stories
59
There is still a difference, however, between this emotional and cognitive activity and the external actions we take in a game or electronic
narrative. Perhaps the most important difference is that in the latter
we are conscious of our activity, which changes our relation to the
story. (Murray, 1997, p. 294)
First Murray says that the difference is in the taking of action. But
then she points out that we are conscious of our activity in ‘electronic narrative’, which we supposedly are not of our activities
when reading or viewing a traditional narrative. There are at least
three lines of criticism to this: First, what kind of ‘actions’ are we
‘taking’ in relation to the narrative? This point is unclear. Second, if
an ‘electronic narrative’ is working well (that is, according to the
aesthetic criteria Murray hints at in other places in the book), is it
not because the audience performs most of its activity unconsciously? Finally, is not reading or watching a traditional narrative
also conscious? Do we not ponder what will happen next, who a
mysterious character is, and why things turned out the way they
did, all on a conscious level? (See, e.g., Trabasso & Magliano, 1996.)
In summary, Murray does not present clear criteria for categorising
stories into participatory and non-participatory, but she advocates a
difference, closely tied to the computer, judged by her choice of the
term ‘electronic narrative’.
4.2.1.3 Ryan’s demarcation
Ryan classifies the ‘new forms of discourse and literary genres born
out of the computer’ (Ryan, 1999, p. 2) in the three main categories
of ‘Computer as (co-)author’, ‘The computer as medium of transmission’, and ‘The computer as theater’:
1. Computer as (co-)author
1a. AI story generation
1b. Eliza
1c. Non-meaningful processing of words in experimental literature
2. The computer as medium of transmission
2a. Digitized print texts
60
Participating in a story
2b. E-mail and Usenet
2c. Tree fiction and collaborative literature (not hypertext)
2d. Electronic serials (soaps)
3. The computer as theater
3a. Hypertext
3b. M U D s and M O O s
3c. Interactive drama (virtual reality)
3d. Computer games (including Pac-Man, Zork, and Myst)
3e. C D -R O M multimedia works (children’s living books, artist’s
works)
3f. Computer-modulated texts (poetry machines, cybertexts)
However, this classification has some odd features. Why is transferring an existing print text, such as a book, into a digital form, such
as a C D -R O M , creating ‘new forms of discourse and literary genres’?
Is it not still the same thing no matter whether it is printed or presented digitally? It is only the third of these categories that resemble
the concept of participatory stories as used in this book. As will be
discussed later, categories 3b, 3c, and parts of 3d are included of the
concept of participatory stories.
The final approach to be considered is Aarseth’s demarcation, on
which the demarcation used in this book will be based.
4.2.1.4 Aarseth’s demarcation
Aarseth’s (1997) book Cybertext: perspectives on ergodic literature—
which builds on ideas from Aarseth (1994)—can be seen as an entry
into the debate about differences between old and new media.
However, Aarseth does not draw his border between printed and
electronic media. Instead, his analysis focuses on structures in the
media, resulting in the two categories ergodic texts and nonergodic
texts. Ergodic texts are those which require special physical activities on the part of the reader or user. These two categories do not fit
the categories old/printed and new/electronic. Indeed, Aarseth
points out that there are old printed works of ergodic literature as
well as new electronic works of nonergodic literature. Aarseth’s
book is one of the most systematic and clearly laid-out accounts of
Participatory stories
61
the textuality of the new media. Aarseth shows great awareness of
the history of the texts and what it is like to read (or use) them.
The categories of ergodic and nonergodic are useful, but there is a
problem with Aarseth’s way of defining the terms, or rather, the
features he uses to separate them, that is, their distinguishing features. Especially, criticism will be offered on the notion that physical actions distinguish ergodic works from nonergodic ones. It will
be argued that ‘physical actions’ should be replaced by ‘mental actions’. This will move the focus from the work to the reader or
user.
Before coming to the criticism, let us look closer at Aarseth’s
concept of ergodic text. In his discussion, Aarseth introduces the
concept of cybertext, which together with hypertext (in its standard
meaning) makes up the larger concept of ergodic text. To put it another way, cybertext and hypertext are subsets of ergodic text.
Aarseth sometimes mixes the usage of the terms cybertext and ergodic in his book, so for the purposes of this section, cybertext can
be taken to mean the same thing as ergodic text. Although there is
no definition in the classical sense of ergodic text or cybertext in his
book, this is what comes closest to it:
The concept of cybertext [as well as ergodic text] focuses on the mechanical organization of the text, by positing the intricacies of the
medium as an integral part of the literary exchange. However, it also
centers attention on the consumer, or user, of the text, as a more integrated figure than even reader-response theorists would claim. The
performance of their reader takes place all in his head, while the user
of cybertext also performs in an extranoematic sense. During the cybertextual process, the user will have effectuated a semiotic sequence,
and this selective movement is a work of physical construction that
the various concepts of ‘reading’ do not account for. This phenomenon I call ergodic, using a term appropriated from physics that derives
from the Greek words ergon and hodos, meaning ‘work’ and ‘path.’ In
ergodic literature, nontrivial effort is required to allow the reader to
traverse the text. If ergodic literature is to make sense as a concept,
there must also be nonergodic literature, where the effort to traverse
the text is trivial, with no extranoematic responsibilities placed on the
reader except (for example) eye movement and the periodic or arbitrary turning of pages. (Aarseth, 1997, pp. 1–2, italics Aarseth’s)
62
Participating in a story
Some examples of ergodic literature that Aarseth mentions are the I
Ching (an old Chinese ritual text), Composition no. 1 by Marc
Saporta (a ‘novel’ consisting of a box of loose pages to be shuffled in
any order), Adventure by William Crowther and Don Woods (a text
adventure game), and Afternoon by Michael Joyce (a digital hypertext).
On the difference between ergodic and nonergodic, Aarseth
writes: ‘If we are to define this difference as a dichotomy . . ., it
would have to be located within the work rather than within the
user’ (Aarseth, 1997, p. 179). So, in Aarseth’s view, ergodicity is a
feature of the (literary) work itself, and this feature requires the user
to carry out physical activities that are not necessary with nonergodic texts.
The problem with Aarseth’s argument is that the features he uses
to distinguish ergodic texts from nonergodic texts are not sufficient
to do the job. He states the central role of physical actions by saying
that ‘the user of cybertext [and ergodic text] also performs in an
extranoematic sense’ (Aarseth, 1997, p. 1) and in ‘nonergodic literature, . . . the effort to traverse the text is trivial, with no extranoematic responsibilities placed on the reader except (for example) eye
movement and the periodic or arbitrary turning of pages’ (Aarseth,
1997, pp. 1–2). But do ergodic texts really require physical actions
that nonergodic texts do not? From a closer look, it appears that
‘periodic or arbitrary turning of pages’ is not only a feature of nonergodic texts, but of ergodic ones as well. For example, the works
Pale fire, Composition no. 1, and Hopscotch, which Aarseth gives as
examples of ergodic texts, require no more physical action than this.
They are printed works and require no more than ‘periodic or arbitrary turning of pages’. So, this feature cannot be used to separate
ergodic from nonergodic. Further, does ‘no extranoematic responsibilities . . . except . . . eye movement’ distinguish nonergodic
works? No, there are examples of gaze tracking technology used to
control the unfolding of a drama in interactive multimedia (e.g.,
Hansen, Andersen, & Roed, 1995), clearly an example of an ergodic
work. It appears that neither mere turning of pages nor eye movement distinguishes ergodic texts from nonergodic texts. These are
Participatory stories
63
the only examples of physical actions supplied by Aarseth in the
book.
If the argument is correct, then it follows that ‘physical actions’
is not a sufficient feature to distinguish ergodic texts from nonergodic texts. One could try to uphold the distinction by imagining
other features, such as that only ergodic works require the pushing
of buttons or otherwise manipulating a special user interface of the
work. This does not help, however, because it is always possible to
transfer a nonergodic work into the same technology that normally
supports ergodic works. For instance, one could digitise a traditional, nonergodic novel and put it into a computer, but that would
not turn it into an ergodic text. The argument is that there is no set
of features that focuses on physical actions which will be sufficient
to distinguish between ergodic and nonergodic, at least not which
will be meaningful as a distinction. Instead, ‘mental actions’ can
replace Aarseth’s criterion, as it is arguably within the user’s head
that the activity actually takes place.
The result of the above discussion questions the idea that the difference between ergodic and nonergodic lies only in the work. If
ergodic works partly require a different type of thinking, which
systematically differs from the thinking required by nonergodic
works, it would be fair to say that the difference is also in the user.
So, should we abandon the term ergodic altogether? No, the concepts of ergodic and nonergodic are still useful, but, as has been discussed, the distinguishing features are neither a difference in the
physical actions required by the user, nor located only within the
work itself. Rather, the difference is the mental activities a user carries out. To transfer the difference between ergodic and nonergodic
from the physical to the mental does not go against Aarseth’s original intention, which can be seen as showing that electronic text and
printed text are simply inappropriate concepts to use when talking
about differences between about new and old media. In other
words, the criticism here has not changed which texts fall under
ergodic and nonergodic, but only supplied other means by which
the concepts may be separated. Taking this critique to its limit
would mean that a mental perspective is necessary in the analysis of
64
Participating in a story
the difference between ergodic and nonergodic literature. This supports the strong version of the cognitive thesis, as formulated in this
book.
Another approach which bases the distinction of traditional stories and participatory stories on physical actions is put forward by
Wilhelmsson (2001). Wilhelmsson views physical, bodily action as
central to computer games, and also as the feature that distinguishes
computer games from, for example, movies. However, he concedes
that some games, such as Zork (classified as interactive fiction—see,
e.g., Costanzo, 1986; Aarseth, 1997; Murray, 1997—in the categorisation used in the present book) is only loosely connected through
such a ‘tactile motor/kinesthetic link’. Even if it is reasonable to
marginalise interactive fiction in a theory of computer games, it is
not reasonable to do so in a theory of participatory stories. On the
contrary, in a theory of participatory stories, interactive fiction is a
prototypical instance. Wilhelmsson’s suggestion may hold for many
computer games, but it cannot be used as a general distinction between participatory stories and non-participatory stories, for the
same reason discussed above in relation to Aarseth’s approach.
4.2.1.5 The present demarcation
The demarcation of participatory stories used in this book builds on
Aarseth (1997), with some extensions. Since Aarseth has a literary
studies perspective, he mainly considers (verbal) texts, and consequently excludes other media from discussion. The membership of
some phenomena is unspecified in his approach. He does not categorise children’s pretence or live role-playing games. In the present
book, the concept of participatory stories also encompasses these
instances, since they are argued to involve the same general cognitive ability. It is unclear whether Aarseth considers all computer
games to be cybertexts (probably he does not, on the same grounds
that is given below to exclude some instances of computer games
from the category of participatory stories). The extensions of the
categories are shown as sets in Figure 4.2.
Figure 4.2 is rather complex, so let us consider what it says in
some detail. The two main categories are participatory stories and
Participatory stories
65
non-participatory stories. Participatory stories include pretence, live
role-playing games, board role-playing games, most of Aarseth’s
concept cybertext, some computer games, interactive fiction, and
adventure games. The categories computer games and interactive
fiction partly overlap; there are computer games which are not interactive fiction, and there is interactive fiction that is not computer
games (such as so called puzzle-less interactive fiction, e.g., The space
under the window by Andrew Plotkin). There are also computer
games that are not considered participatory stories (these are games
with very little stimuli which could generate a narrative response,
e.g., abstract games14 such as Tetris (Pazhitnov, 1985)). Adventure
games are a type of interactive fiction computer games. Some parts
of Aarseth’s category cybertext are excluded, for example, automatic story generators which do not involve the audience. The
category of participatory stories excludes Aarseth’s (and the general
notion of) hypertext. Why are hypertext stories outside of the concept of participatory stories? Hypertext consists of texts and links.
Links make reference to other (parts of) texts. But in participatory
stories, such as interactive fiction, there is no question at all of references to other texts—there are consequences of the user’s actions
upon the story and the simulated story world. Hypertext is not
procedural (Murray, 1997). Comprehension of hypertext is not like
comprehension of participatory stories. Following links in hypertext is a navigational task, often performed without cues, and puts
load on working-memory in that one needs to memorise part of the
way one is taking in the navigational structure (Wenger & Payne,
1996). The difference between hypertext and participatory stories
will be further discussed in the section ‘What does it mean to participate in a story?’.
14
Tetris is abstract in the sense that it contains geometrical figures, and not characters
and natural objects. In another sense, Tetris is a very concrete game with a simple
objective and a limited set of motor actions on part of the player.
66
Participating in a story
participatory stories
pretence
live role-playing games
Aarseth’s ergodic literature
Aarseth’s
cybertext
non-participatory
stories
computer games
interactive fiction
fairytales
novels
short stories
plays
adventure games
cinema
everyday narrative
board role-playing games
Aarseth’s & others’ hypertext
Figure 4.2. The classification of participatory stories and non-participatory stories as
used in the present book, shown using sets. Aarseth’s (1997) classification is shown
for comparison.
The other main category is non-participatory stories, or traditional
stories. Here we find, for example, fairytales, novels, short stories,
plays, cinema, and everyday narrative.
The granularity of the present categorisation is in one sense very
coarse, in that there is a dichotomy between participatory stories
and non-participatory stories. Of course, it is relevant to ask
whether it is reasonable to have two major categories. Are the instances within the category of participatory stories similar enough
to warrant such a categorisation? Despite differences in appearance
among the individual participatory stories in terms of text, sound,
Participatory stories
67
video, and different physical ways in which to control them, the
argument in the following sections is that it makes sense to apply
such a generalisation. The argument, which will be expanded later,
is basically that story participation is best defined as a cognitive process, independent of physical circumstances and sense modality.
The term ‘participatory stories’ was selected for the phenomena
studied in this book. ‘Participatory’ is intended to mean the same
thing as ‘interactive’, but it was chosen because it is not as over-used
and is less loaded compared to the term ‘interactive’ (see Aarseth,
1997, for a discussion). ‘Participatory’ also leads associations away
from the computer, which goes well with the broad view of participatory stories in this book (live role-playing, which has nothing to
do with computers, is also an example of participatory stories in
this classification).
Before coming to a definition of what it means to participate in a
story, there is a need to consider two related factors that are often
mentioned in the context of participatory stories: fidelity and multimodality.
4.2.2 The role of fidelity and multimodality
The issues of fidelity and multimodality are thought by some researchers to be central to the field of participatory stories, as will be
seen later in this section. Fidelity is used to mean how accurately
something is represented. In a computer game, for example, a player
may be represented by a crude square, a line-drawing of a human, or
a detailed three-dimensional figure with full facial features—three
versions with increasing fidelity. Multimodality here means involving more than one sense modality at once. For convenience, one can
think of modalities as the basic five senses of vision, hearing, touch,
smell, and taste (although there are arguably others, e.g., pain, balance, and proprioception). Multimodality could be present either in
terms of participation—interactivity—the manner in which the audience influences the story (multimodal input), or it could be in
terms of how the participatory story presents information to the
audience (multimodal output).
68
Participating in a story
Are these features really fundamentally connected to participatory stories? The case for fidelity and multimodality in participatory stories will now be reviewed, looking at examples of participatory stories, considering arguments from researchers in the field,
and assessing psychological theories which may support such a position. The final conclusion will be that fidelity and multimodality
are in principle unrelated to participatory stories, and this is formulated as the modality-independence assumption of participatory stories.
4.2.2.1 Fidelity and multimodal output
Let us first consider the role of fidelity and multimodal output in
participatory stories. The most common form of multimodality in
participatory stories would be the presentation of the story to the
audience utilising the senses of vision and hearing when presenting
audio and video material.
Scholars discussing participatory stories often dwell on the feature of multimodality. Murray (1997) does not seem to be able to
make up her mind about whether a multisensory interface is important to participatory stories or not. When discussing ‘immersion’ as
a key property of the new medium, she describes it as ‘the sensation
of being surrounded by a completely other reality . . . that takes
over all of our attention, our whole perceptual apparatus’ (p. 98).
Thus, Murray attaches importance to being able to experience the
fictional world using as many senses as possible. She discusses the
three-dimensional Sony I M A X Theater (non-participatory like traditional cinema), and says: ‘The size of the film means an increase in
information, offering a richer and therefore more persuasive visual
illusion. It is not merely a larger image but a more present reality’
(p. 45). Also, in relation to computer games, Murray expresses a
view of ‘more is better’—she calls the shift from text-based interfaces to graphic-based ones (or at least ones with spatiality) ‘progress’ (p. 190), so apparently, she views the latter more positive than
the former. But when discussing M U D s and the adventure game
Zork (both consisting of text only), Murray notes that multisensory
interaction is not apparently necessary to obtain a successful experience in a participatory story, as the success of these works shows: ‘It
Participatory stories
69
demonstrates that the potential for compelling stories does not depend on high-tech animation or expensively produced video footage
but on the shaping of such dramatic moments’ (p. 53), and ‘narrative beauty is independent of medium’ (p. 273).
Laurel (1993) also stresses the importance of multimodality for
immersion in interactive media: ‘Tight linkage between visual, kinaesthetic, and auditory modalities is the key to the sense of immersion that is created by many computer games, simulations, and virtual-reality systems’ (p. 161). However, it is unclear whether Laurel
specifically addresses qualities of experiencing a story, as opposed to
having any experience of a surrounding.
It is easy to be swept away by the thought that a higher degree of
fidelity and multimodality, such as higher screen resolutions, more
colours, force-feedback hand-controllers, and five-channel surround
sound, are essential in creating a truly participatory story. But one
must distinguish between commercial factors such as what types of
games sell the best, and what the true possibilities of interactive media are (for discussions of this from a game design perspective see,
e.g., Crawford, 1993; 1994). Besides marketing reasons, the underlying, implicit argument in claims of ‘more is better’ is this: rewarding
participatory stories should be multimodal because real life is multimodal. However, this argument is faulty. Drama is not like real
life. In drama, things are condensed and intensified, all to maximise
the audience’s experience. Consider instead the typical storytelling
situation in a historical perspective. The most common situation
has undoubtedly been when a storyteller captures the imagination
of an audience by telling a story using words only. Other common
rewarding story experiencing situations include the novel, such as
War and peace (no pictures). Multimodality is not central to participatory stories because a story is comprehended at a more abstract
level, which is separate from how it is represented.
Multimodality may be present in many participatory stories, but
it cannot be used as a feature to tell participatory stories apart from
non-participatory stories. There are many examples of works,
which we would consider to be typical participatory stories, that
involve textual input and output only, for instance, text adventure
70
Participating in a story
games. Therefore, multimodality cannot be a necessary feature of
participatory stories. It is even highly questionable whether
multimodality at all adds to the experience of immersion in stories
(see Gander, 1999 for an extended argument and overview of the
literature).
Is there psychological evidence that multimodality is important
for story comprehension? Let us first turn to laboratory research on
memory. Paivio’s (1991) dual-coding hypothesis suggests that multimodality may affect comprehension. If some content is presented
as both pictures and verbal material this may result in two codes in
memory, which leads to a higher degree of recall because there are
more retrieval cues. But this holds only if the same information is
presented in both modalities. The usual case for participatory stories would be that different information is presented simultaneously
in several sense modalities. Thus, the dual-coding hypothesis would
play a small role in these cases. Even if recall were facilitated by two
codes, what does this tell us about experiencing a story? The dependent variable in most memory studies is memory performance,
usually on lists of words, or sometimes single actions in isolation. It
is very unclear whether research on unrelated words can be generalised to story comprehension, which involves larger units of meaningful material.
How is the case for multimodality in studies dealing specifically
with story comprehension? Contents of news stories from print and
television are remembered equally well by adults (Furnham, De
Siena, & Gunter, 2002). Story comprehension seems to work in the
same way regardless of whether the story is presented audio-visually
or as text, for predictive inferences (Magliano, Dijkstra, & Zwaan,
1996), and comprehension of events (Magliano, Miller, & Zwaan,
2001): ‘The pattern of results was similar for narrative film as they
are for narrative text. This finding suggests that there are general
mechanisms for event understanding that operate independently of
medium or mode of experience’ (Magliano, Miller, & Zwaan, 2001,
p. 533). And: ‘This finding bolsters the claim that the higher-level
processes involved in situation understanding are generalisable
across experiences, whether they occur in text, film, or in real life’
Participatory stories
MEDIUM
COGNITIVE SYSTEM
Sense-modality and
genre specific perception
and interpretation
Movie
71
Sense-modality
and genre free
interpretation
Cinematic conventions, etc.
Story:
Book
Speech
Literary
conventions, etc.
events, goals,
plans, causation,
etc.
Basic listening
competence
Figure 4.3. A schematic view of the sense-modality independence of story understanding.
(Magliano, Miller, & Zwaan, 2001, p. 543). On the whole, the psychological evidence points to the conclusion that the medium does
not matter when it comes to story comprehension. Figure 4.3 illustrates the idea of a modality-independent view of story comprehension.
4.2.2.2 Multimodal input
Do participatory stories depend on the manner in which participation is performed, for instance, typing an action as a textual command compared to motorically carrying out an action? That is, is
multimodal input an important consideration when theorising
about participatory stories?
Aarseth (1997) and Wilhelmsson (2001), emphasise the importance of physical, motor actions when describing the essence of participatory stories. Wilhelmsson also expresses the view that the visual modality (and to some extent also the auditive modality) is an
72
Participating in a story
integral part of participatory stories, because motor actions are
linked to visual results in the game. The view that the body is involved in cognitive activity has become popular in cognitive science
and related fields (Wilson, 2002). Wilhelmsson makes a strong claim
when he states that ‘sensorimotor experiences are what structure
subjective experiences’ (Wilhelmsson, 2001, p. 149) and pays homage to researchers such as Lakoff, Johnson, C. Johnson, Grady, Narayanan, Fauconnier, and Turner.
In an attempt to capture the general features of computer games,
Wilhelmsson (2001) refers not to specific concepts such as Avatar or
player character, but to ‘visual container’ and ‘sound-scape’ (Wilhelmsson, 2001, p. 150). However, in the use of these terms, he implies that there is generally something visual about computer games
and that they inherently have sound. This is not the case with many
computer games and participatory stories, such as Anchorhead
which was used in the empirical studies in the present book. Further, Wilhelmsson draws heavily on Gibson’s ecological theory of
perception when discussing features of computer games, such as
space. However, Gibson’s approach may only be applicable for
cases where there is a visual field. When considering computer
games without a visual field, there can be only metaphorical uses of
Gibson’s approach, through mental representations, and this is
clearly not how it ought to be understood.
There seem to be two ways in which one may attempt to rescue
the view that motor-kinesthetic features are fundamental to computer games in general, given that also interactive fiction (e.g., Anchorhead)—which does not have direct control and does not include
any visual features beyond text—is a kind of computer game.
First, one may say that in playing interactive fiction, there are no
visual or motor actions directly related to the game, but via mental
representations, the actions in the game are translated into a motorkinesthetic code, in the end creating the same kind of subjective
experience that would be obtained with a visual computer game of
direct control. But why would textual interactive fiction be translated into motor-kinesthetic actions of manipulating the interface
(such as those which are available for computer games controlled
Participatory stories
73
with, e.g., a joystick)? And even if one held this view, one has already rejected the importance of motor-kinesthetic features and
accepted the importance of the mental dimension, in claiming that
mental representations of the participatory story are formed.
Second, one may say that motor actions are not integral to computer games but that they are merely circumstantial. Agency may
be instantiated in many forms, such as motor action in direct control, through keyboard commands, via speech commands, or
merely by directing one’s gaze to various locations. Then, any of
these instantiations would be translated into a more general, conceptual—mental—level where the player is ‘doing things in the
game’. The experience of playing computer games (and we would
say, participating in any story) would then be an experience based
on a mental process rather than on a motor-kinesthetic process.
There is psychological research which investigates the impact on
memory of motor actions. Within the enactment paradigm, sometimes called subject performed task paradigm, results show that actions actually performed by participants are better remembered
than actions that are merely read or heard (Nilsson, 2000). Although there are conflicting results within the field, the explanation
for the effect seems to be that the action facilitates encoding of the
verb and the noun in the action, but the action is represented similarly in memory regardless of whether it was read, heard, or enacted—there does not seem to be a motor component in the memory trace (Nilsson, 2000) (see also the note in the last section on
generalisation from memories of single unrelated actions to stories).
4.2.2.3 Modality-independence assumption of participatory stories
From the above discussion the following conclusions are drawn:
(i) there are examples of prototypical participatory stories which do
not utilise multimodal input or output, (ii) psychological research
on traditional story comprehension shows that stimuli in various
modalities are processed in a similar way, (iii) psychological research
on action shows that there does not seem to be a motor component
in memory of enacted actions compared to actions that were merely
imagined or observed. This leads us to state the modality-
74
Participating in a story
independence assumption of participatory stories: Multimodality is not
a necessary criterion in order to maintain a difference between the
classes of participatory stories and non-participatory stories. That is,
still images, moving images, text, and sound, as well as the features
of the senses used for reception, such as vision, hearing, and touch,
are all ultimately not an essential part of the concept of participatory stories. Nor has the range of input devices used, such as joysticks, virtual-reality goggles, eye and gesture tracking devices, any
relevance. Rather, the concept of participatory stories can, and
should be, generalised from its manifestation in discourse (that is,
how it actually appears in any media), and how interaction is
achieved. Dwelling on technical issues of multimodal interaction
will only confuse the investigation and lead away from the core of
the phenomenon of participatory stories.
The assumption that the cognition of story understanding is
similar across media in participatory stories has methodological implications. If the assumption holds, it gives us greater possibility to
generalise from empirical findings obtained in the study of one type
of participatory story, for example, text adventure games, to other
types, such as graphical adventure games, and beyond.
Having concluded the independence of multimodality from the
concept of participatory stories, the discussion can now proceed to
the heart of the matter: What does story participation really mean?
4.2.3 What does it mean to participate in a story?
It was claimed above that it is meaningful to divide individual story
works into the categories participatory and non-participatory. In
order to give substance to such a division, we need to specify in detail what it means to participate in a story. As was discussed earlier,
the term ‘participatory’ is similar to the term ‘interactive’, which is
more commonly used when talking about these issues. In this section, these terms will be used interchangeably. First, a series of suggestions as to what participation might be will be considered, arguing against these as characterisations of real participation, and finally
arriving at a definition of story participation used in this book.
Participatory stories
75
4.2.3.1 Approaches to participation
Interaction can generally be viewed as information exchange between at least two systems. Concerning stories, interaction means
that there must be some kind of information exchange both to the
audience from the story, and from the audience to the story: a feedback loop, influencing both audience and story. However, this
characterisation is very vague and places only mild constraints on
what participation might be in the context of participatory stories, a
fact that has generated several views on the matter.
It is possible to claim that all storytelling is participatory, using at
least two separate arguments. First, if meaning is created in interaction between text and reader, does this not mean that all stories are
therefore participatory? This argument has already been countered
above and will not be further discussed here. The second argument
is that in stories (at least in some of them), the audience imagines
the story world and experiences themselves as being ‘inside’ the
story and taking part in it. Therefore, all stories are participatory.
There is no reason to doubt that these experiences are real, but they
are not examples of true story participation because they fail to satisfy the general requirement of interaction above: no information is
fed back from the audience to the story.
Another argument claims that perhaps not all, but much of storytelling is participatory because of the social nature of storytelling.
When observing everyday storytelling, such as people gathering
around a dinner table and talking about the day’s events, one notices that it is common that several people take part in the telling of
a story (Norrick, 2000). Someone may begin to tell a story, which is
then filled in, embellished, or rejected by the utterances of other
people. Is this a case of participatory stories? A closer analysis shows
that it is rather the shaping of the telling of the story that is participatory. It is not a case where the audience participates and changes
the events and the outcome of the story.15
Social situations which would be included in the concept of participatory stories,
as discussed above, are the activities of board and live role-playing. However, even
15
76
Participating in a story
There is another argument according to which any book or
movie is participatory. It is possible to read the pages of a novel in
an order different from the order in which they are numbered. That
way the unusual situation may occur, for instance, that the hero of
the story first is dead and then comes alive. With a video cassette
recorder, the audience can achieve similar effects with a movie by
interacting through controlling playback: stopping, rewinding, and
using fast forward.16 Yet a typical novel and a typical movie do not
belong to the category of participatory stories, for two reasons.
First, even though it is possible for the audience to use them in the
ways described, it is not the intended way. As was discussed in
Chapter 2, there are conventions for how various storytelling media
should be interpreted. Going too far beyond these conventions,
such as reading the pages of a typical novel by arbitrarily jumping
from one page to another, makes it something other than a traditional
novel. As was discussed in the section on ergodic literature, although it consists of text printed on paper, it is no longer a linear
text, but rather a hypertext, in this case. The second reason why
interaction in the choice of how a story is presented is not participation in the sense intended in this book, is that it is merely changing
the presentation of the story, and not the order of the events themselves or the contents of the events themselves. In the terminology
of Chatman (1993), it is possible to change the discourse while keeping the story intact. For instance, a traditional movie may start by
showing the end of the story, and the rest of the movie is really a
flashback of the main character. An alternative version of the movie
may start at the beginning of the story and end with the main character shown in the storytelling situation. In these two examples the
discourse is different but the story remains the same.
Two strong candidates for story participation will now be discussed: hypertext stories and multiple viewpoint stories. Both are
though these situations could theoretically be carried out around a dinner table, they
are highly specialised and do not occur in everyday storytelling.
16
Of course, the same goes for several other audiovisual media, such as digital versatile discs (D V D s).
Participatory stories
77
examples of interactive media, but are they cases of story participation?
Is story participation of the same kind in hypertext stories, compared to participatory stories? If we compare the user interaction in
a hypertext story, such as Afternoon (Joyce, 1987), to an adventure
game, such as Deadline (Blank, 1982), we find a major difference. In
Afternoon, which is about a man who may have witnessed the car
wreck of his former wife and the death of his son, it is again the
order of presentation (Chatman’s discourse) that is influenced by
our participation, but not the story itself. The audience can learn
new things about the story by using it several times and following
the links in different order, but it cannot change the basic storyline.
There is no flow of information from the audience to the story. In
contrast, Deadline, which is about investigating a supposed suicideturned-into-murder, the outcome of the story itself depends on
which actions the audience performs. Whether the murder will be
solved, whether there will be more murders, and whether the player
character will survive to the end of the investigation, is all up to the
audience of Deadline.
Another case where there seems to be story participation is in
multiple viewpoint stories. Here, the audience can choose from
whose character’s perspective the story should be told. Murray
(1997) discusses ‘Mobile Viewer Movies’, and talks about the ability
to ‘branch through a story’. But that cannot be what she really
means, because all she discusses is the ability to ‘choose who to follow’ in a given story. As can be seen from the earlier discussion,
again, that is not a branching on the story level; that is branching on
the discourse level, that is, a difference in how the story is presented,
but not how it is constituted.
Murray (1997) holds the feature of agency as central to stories of
the future. By agency she means the feeling of being able to
influence elements of the story. Her concept appears to be a promising component on which to build story participation. However, the
concept is vague in two ways. First, Murray does not say whether it
is necessary that there is an actual influence in order for a medium
or story to have the property of agency, or if perceived agency—an
78
Participating in a story
illusion of agency—is sufficient. Second, Murray is not specific as to
what is influenced. Does a story have agency if the audience, for
instance, is able to change the colour of the protagonist’s clothes, or
is a more fundamental influence required? If so, what kind of influence?
The issue of authorship is sometimes discussed in relation to interaction. Does interaction mean that the interactors in a story become authors of that story? Buckles (1985) argued that the players
of adventure games became authors. However, this seems to be an
illusion resulting from the fact that the game Adventure which
Buckles studied was textual. In Adventure, the player writes text, so
she appears to become an author. But if we consider a version of the
same game (which is isomorphic), converted into a graphical pointand-click interface, we see that textual input is not needed, making
it seem less plausible to call the player an author. Proponents of
hypertext often claim that interactivity enables readers to free
themselves from the less powerful situation as readers and become
authors, with political implications (Landow, 1997). However,
when looking more closely at what happens with participatory stories as well as with hypertext stories, this seems not to be the case.
Instead, the audience interacts with a structure created by someone
else, a designer or author. Murray (1997) is one who differentiates
authors from interactors: ‘The interactor is not the author of the
digital narrative, although the interactor can experience one of the
most exciting aspects of artistic creation—the thrill of exerting
power over enticing and plastic materials. This is not authorship
but agency’ (Murray, 1997, p. 153).17
Although not considered by Murray, perhaps a way to see interactors in ‘digital
narrative’ along the lines of traditional drama, might be more like actors than like
audience, with the difference that they can influence the story.
17
Participatory stories
79
4.2.3.2 The present definition of story participation
Having discussed and criticised various attempts to define what it
means to participate in a story, the definition used in this book can
now be presented:
Participation in a story means perceiving at least two different potential event sequences of a story, and causing one of them to occur by
controlling the actions of an agent in the story.
This definition of participation has several consequences.
For a cognitive system (such as a person) to participate in a story,
potential event sequences must be able to be perceived and be
judged concerning similarity. If there is no way to compare them,
they could be one and the same event sequence (in which case we
could be dealing with a story from a traditional novel, and we
would not say that it is participatory).
Besides perceiving the event sequences, the one who participates
in a story needs to cause one of them to occur. Here we are faced
with a problem. How can the audience, part of the real world,
influence a story world, which is fictive? The audience (one or more
persons) is not part of the story. To perform actions in the story, the
audience needs an agent that is part of the story world. Luckily, as
the concept of story was defined in Chapter 2, an agent is an essential component. By controlling an agent, actions can be performed
on the story world, which in turn causes one event sequence to occur rather than another. The agent is sometimes called ‘player character’ (in adventure games). Even if no player character is assumed
(the audience is ‘playing themselves’) there is some representation of
the audience in the story. If the agent is hindered (imagine a story
where the player character is being tied up by some kidnappers), the
audience also (temporarily) loses the ability to influence the game. If
the player character dies, the possibility to interact with the story
also ceases (usually). This view of an agent is similar to Wilhelmsson’s (2001) ‘Game Ego’, proposed as a component of computer games in general: ‘The Game Ego is an extension of the human body container. The Game Ego is a function that affords and
80
Participating in a story
exerts action onto and within the game environment’ (Wilhelmsson, 2001, p. 247). The audience may try to force themselves
onto the story and try to change it, such as modifying the software
or hardware which runs the participatory story (i.e., we would call
it ‘cheating’). Although this is some sense is participation (with the
hardware or the program), it would not be story participation. Another similar case is when a person flips the pages of a traditional
book novel open at various places and reads. For the same reasons,
this is not story participation because it is not participation with the
story, which can only be through an agent, but with the book as a
physical object.
Another consequence is that the interaction needs to be brought
about through a choice. It is not sufficient for a story to change as a
result of mindless, random interaction. Neither is it sufficient to
influence how the story is presented, Chatman’s discourse level, but
the influence is on Chatman’s story level.
According to the definition, no physical manifestations of participation of any kind are necessary. Consequently, one could participate in a story all in one’s head. This could be a weakness or a
strength of the definition, depending on how liberal one wants to be
concerning the concept of story participation. Two cases of purely
mental story participation that we perhaps would want to include,
and that are now included, are the activities of dreaming during
sleep and having fantasies while awake.
Another consequence of the definition is that it does not address
the fictional status of the story. This means that real world situations qualify as story participation. An example of story participation would be the situation of you sitting at a restaurant ordering
food. You are participating in your life’s story.
Note that changing the sequence of events demands a way for the
audience to influence, but the manner in which this is done is left
unspecified, in accordance with the modality-independence assumption made earlier (be it by pulling a joystick, by typing text on a
keyboard, by changing one’s gaze using an eye-tracking system, or
through some other means).
Participatory stories
81
One problem with the definition that seems inevitable is that it
allows for very trivial cases of choice, that is, the potential event
sequences may differ only slightly and in a trivial way. For example,
we are forced to call it story participation even if the only choice in
the entire story is between moving a piece of dust on the floor and
letting it stay where it is. Not a very exciting example of story participation.
The definition here is similar to other approaches, though somewhat more restrictive about what is considered participation. For
instance, it excludes the mere ability to choose viewpoint, as discussed above.
4.2.3.3 Definition of participatory stories
Now participatory stories can be defined by building on the definition of story participation:
A participatory story is a physical system separate from the audience
allowing participation in a fictive story.
Let us look at the features making up a participatory story.
In Chapter 2, story was defined as a mental entity. However, as
can be seen in the definition, participatory stories are not mental,
but refer to physical systems with which the audience is interacting.
This allows us to talk about individual works, such as computer
games, as participatory stories. It follows from this that participatory stories are not stories themselves. Instead, they are the physical
stimuli which generate narrative response. In some cases, these systems are computer hardware running the appropriate software. But
it can also consist of the collection of people, props, and environment as in live role-playing games.
Another requirement made for the same reason is that a participatory story should be separate from the audience. This excludes
cases of participatory stories completely inside the head of a person,
but it adds the convenience of being able to talk about participatory
stories and by this mean separate, physical works, such as computer
games.
82
Participating in a story
It is only the possibility of participation that is necessary for
something to be called a participatory story. If we were to require
actual participation, then we would have the peculiar case when the
same physical system is not, and then suddenly, becomes a participatory story, as the audience’s activity changes over time. What happens in an actual situation depends on the audience; it may do nothing, and even though the experience probably will be less full, we
will still call such a system a participatory story.
Finally, the notion of fiction is integrated into the concept of participatory stories. Participatory stories are about fictive worlds, not
the actual real world. This last point may need some clarification,
which brings us to the fourth question that challenges any theory of
participatory stories: the question of whether participatory stories
are fictional.
4.2.4 Are participatory stories fictional?
The question of whether participatory stories are fictional or not is
interesting for several reasons. Compared to non-participatory stories, such as novels and movies, participatory stories not only represent, but also involve physical actions. How can participatory stories be fictional when the audience carries out actions when experiencing them? The answer to this question may have implications
also for our general notion of fiction. If this notion is based on only
non-participatory stories, the discussion could also influence our
view of what fiction in general is. Researchers such as Aarseth
(1997) and Wilhelmsson (2001) can be interpreted as giving the answer that participatory stories are not fictional. But if participatory
stories are not fictional, then what are they? The proposals of
Aarseth (1997) and Wilhelmsson (2001) will be argued against with
the aim of showing that participatory stories are indeed fictional.
Let us first consider the arguments of Aarseth (1997).
4.2.4.1 Can fiction involve testing causal relationships?
Fiction, to Aarseth (1997), is ‘a portrayal of invented events or
characters, usually in the form of prose (short stories, novels, etc.),
Participatory stories
83
constructed in a way that invites rather than dispels belief.’ (p. 50).
Aarseth notes that the term ‘interactive fiction’ is used in the literature, but that no researcher tells in what sense interactive fiction is
fiction (up until 1997, when Aarseth’s book was published). In interactive fiction ‘the user can explore the simulated world and establish causal relationships between the encountered objects in a way
denied to the readers of Moby Dick’ (p. 50). This is indeed a feature
of interactive fiction, and also of participatory stories in general.
However, Aarseth views action as the stumbling block to fiction: ‘a
fiction that must be tested to be consumed is no longer a pure
fiction; it is a construction of a different kind’ (pp. 50–51). Since the
role of fiction is unclear, Aarseth claims that the term ‘interactive
fiction’ is meaningless and suggests that it should not be used.
There seem, however, to be steps missing in Aarseth’s argument.
There is nothing in his definition of fiction that does not allow exploration and testing of causal relationships. It should be unproblematic to view the characters, objects, places, etc. in a work of interactive fiction to be fictional in Aarseth’s sense; they are invented.
But can a causal relationship be fictional, for example, that you need
to find a certain (fictional) key in order to unlock a (fictional) door?
Since keys unlock doors also in the real world, we may be confused
and regard this as an actual causal relationship that holds in the real
world too. But another example makes it clear that the causal relationship is actually fictional (invented in Aarseth’s terms), for instance that giving cheese to an (anthropomorphic) mouse makes it
whisper the location of a secret door. Here, the causal relationship
is fictional because giving cheese to a real mouse will not have this
effect. Indeed, the causal relationship cannot be real for another
reason: the objects or characters which are causally related are
fictional (invented). How can a real causal relationship hold between
objects that do not exist? The causal relationships tested in a work
of interactive fiction are fictional. (If we do not want to make any
ontological claims, we can say, in parallel to the definition of fiction
from Chapter 3, that the audience will consider that the causal relationship should not be evaluated in the real world.) Furthermore,
causal relationships seem to exist in non-participatory stories as
84
Participating in a story
well, such as in a novel. The actions of fictional characters have effects in the world of the novel. So, fictional causal relationships do
not seem to be such a strange thought.
In summary, Aarseth’s argument that interactive fiction is not
fiction because it contains elements of testing causal relationships is
not valid. Indeed, the view that something can be fiction and still
allow exploration on the part of the audience is consistent with
Aarseth’s definition of fiction, as well as with the definition of fiction in this book (Chapter 3).
4.2.4.2 Do computer games use representations?
Addressing the question of whether participatory stories are
fictional is uncommon in the literature, but Wilhelmsson (2001)
touches on the issue while arguing against Laurel’s use of theatre as a
metaphor for computer games (since Wilhelmsson discusses interactive fiction as computer games, his analysis is relevant to the question of the fictional status of participatory stories):
I would say that there is not representation of something in some
(maybe even most) computer games when it comes to the core of subject matter of a game and its manifestations. Consider Pac-Man for instance. The maze in Pac-Man making up the game board and the game
environment really is a maze. The hunt is a hunt. The actions performed are performed. It is a real hunt, not a fake one. The way it is
performed with a Game Ego is in some respects different from running around being hunted by ghosts in real life (if now there are
ghosts in real life that is). Nevertheless the experience is quite similar
and uses to a large extent the same motor based cognitive schemas.
(Wilhelmsson, 2001, p. 135)
Wilhelmsson is making two claims here: first, that some computer
games (where Pac-Man is an example) do not represent real things—
the ‘things’ in the computer games are what they are by virtue of
their intrinsic properties. This claim is logically separate from his
second claim, which is that the ‘things’ (at least those mentioned)
found in Pac-Man are ‘real’. Now, regardless of which definition of
fiction one uses, the last claim must reasonably be using ‘real’ in
Participatory stories
85
opposition to ‘fictional’. Under this interpretation, the claim is that
Pac-Man is not fictional. The claims are logically separate because a
representation could be either fictional or non-fictional: A movie
may represent some real events, or it may represent some invented,
fictive events. However, there is a one-way dependency between the
two claims: For something to be fictional, it must be a representation (real things are non-fictional). If we agree with Wilhelmsson
that Pac-Man does not contain representations of the things he mentions in the quotation above, then it follows that they cannot be
fictional either.
First, we may question that some things in Pac-Man are not representations. Take the ghosts, for instance. If we consider the ghosts
to be fictional—which would be hard to argue against—then they
are also representations. Indeed, are not all elements in Pac-Man representations? Surely the fluorescent dots of light on the screen represent a maze, much in the same way as a painting of a maze or a
shot of a maze in a movie represent a maze. It looks like a real maze
but it is not. Next, consider actions in Pac-Man. Is the hunt really a
hunt? This depends of course on our definition of hunt. But it seems
reasonable to demand that a real hunt involves real objects (or
agents). Since we do not have that in Pac-Man, it appears strange to
call it a real hunt. What about other actions? In some sense, the audience’s actions have actually been performed, also in the real
world. However, what they mean in the real world is different from
what they mean in Pac-Man. In the game, the action ‘eating a ghost’
may have been performed, but in the real world, the action
amounts to, for instance, pulling the joystick to the right.
Wilhelmsson’s claim that the experience is similar does not add to
the argument, since the experience of a hunt while, for example,
watching a fictional movie—a representation of a hunt—could also
be similar to the experience of a real hunt. Further, the motor actions in Pac-Man do not seem to resemble those in a real hunt: pulling a joystick forward, backward, left, and right (in the game), compared to running by moving your legs (in real life). Even if the actions of the computer game interface were similar to real actions,
86
Participating in a story
this is rather a special case and not something that is either common
or necessary for computer games in general.
The conclusion is that Wilhelmsson presents no convincing argument that Pac-Man (or computer games in general) is real and not
fictional. Indeed, he states a few pages later that ‘the player is able to
activate, through cues from the interface of the computer game, a
simulation of a fictious[sic] world inside the computer’ (p. 145). The
view Wilhelmsson expresses here is consistent with the view of participatory stories held in the present book (under the assumption
that he does not really mean a simulation of a fictitious world, but
rather intends to say ‘a simulation of a world, which is fictitious’).
4.2.4.3 Exactly in what sense are participatory stories fictional?
A claim that participatory stories are fictional must be followed by
an account of how they are fictional. As mentioned before, the main
reason for considering them to be fictional is the observation that
people who experience participatory stories treat what happens in a
participatory story as something that is not real.
A way to go beyond the real/fictional dichotomy may be to introduce a third option between fictional and real, indeterminate. We
could say that the audience is unsure whether the events in a participatory story are real or not. Reality is real, novels are fictional,
and participatory stories are indeterminate. But this is clearly not
what is happening—the audience experiencing a participatory story
is not in doubt. Instead, the audience seems to have exactly the same
relation to a participatory story as to a non-participatory story,
such as a fictional novel—the world they are experiencing is imaginary. Thus, the only reasonable conclusion is to call participatory
stories fictional. However, there is still a need to lay out in detail in
what way they are fictional, considering that the (real) audience carries out (real) actions in participatory stories.
Let us look more closely at what it means to say that participatory stories are fictional, by examining the relation of fiction to the
components of participatory stories. The world in a participatory
story includes, for example, characters, objects, and places. It seems to
be uncontroversial to claim that these elements are fictional in ex-
Participatory stories
87
actly the same way as the story world of a novel is fictional. The
events that take place in the participatory story, for example, that a
king was killed, can also be said to be fictional in the same way.
However, when we come to the actions carried out by the audience,
there seems to be something more than fiction going on. However,
actions carried out by the audience within the story, such as killing
the king, can be said to be fictional. Clearly, no real killing was performed and no real king was involved. (If the audience was questioned in a court of law, it would rightfully deny having killed anyone.)
The action of killing the king (in the story) was performed in
some way by the audience, for example, by pressing a button,
which shot an arrow at the king (in the story). This part of the action is real (pressing a button)—this works since a real body was
used to press a real button. In contrast, the part of the action that
relates to the game world is fictional (shooting an arrow and killing
the king). The part of the action that is carried out by the audience
in the real world—working the interface, such as pressing a button—
can rightfully be said not to be a part of the participatory story
(since it does not contain reference to any elements in the story
world). This amounts to a view of action in relation to participatory
stories as consisting of two domains: a fictional game world domain,
and a real life domain. One nagging question still lingers, with apparent similarities to the mind-body problem as formulated by Descartes: If these are two domains of a different kind, how can they
causally interact? How is it possible for a real audience to influence
a fictional world? The answer is straightforward: The causal influence of the (physical) audience is on (physical) matter, nothing else.
However, the physical matter—such as electrical currents through a
computer system—is the basis of an experience of a game world in the
audience. Speaking of a fictional world as opposed to a real world is
just a convenient shorthand.
The clear distinction between representation and represented
holds nicely in personal computer-based participatory stories. But
what about other participatory stories, such as live role-playing? As
soon the objects, characters, places, events, and actions have a causal
88
Participating in a story
connection to the real world, or when fictional and real objects
share physical substrate, there are effects on the real world as well.
For instance, if a stone is thrown through a closed window during a
live-role playing session, not only did the stone break the window
in the story, but it also did so in the real world. However, this is
not an essential property but is contingent upon the situation that
the story world and real world share physical substrates and actions
are attributed the same meaning in both the story world and the
real world. Throwing a stone at a window in the real world may
represent something else in the participatory story, such as killing a
king (however peculiar this interpretation may be). Even though
there is a causal connection between an action in the fictional world
and the real world, it does not necessarily mean that the actions
carried out in the participatory story world mean that they are also
carried out in the real world.
In conclusion, participatory stories are fictional, unless they are
set up so that the fictional story world has causal connections to the
real world, and the elements of the fictional story world are interpreted in the same way as in the real world. However, this requirement is rarely fulfilled in practice (actually, one can argue that if it
is, we are no longer dealing with a participatory story, but with the
real world). Thus, the conclusion is that participatory stories are
fictional.
In summary, this chapter has presented a demarcation of participatory stories from non-participatory stories, building on a classification of Aarseth (1997) which was extended beyond cases of literature and re-formulated in terms of cognition. Among others, the
category of participatory stories includes adventure games, interactive fiction, some computer games, role-playing games, and
pretence. Typical non-participatory stories are novels, movies,
fairytales, and everyday narrative. Participatory stories were characterised as physical systems separate from the audience allowing the
audience to choose between different potential events sequences in a
fictive story. The choice is made through some influence although
Participatory stories
89
the manner in which the choice is made (in terms of fidelity or
multimodality) was seen as immaterial.
Part I formulated a conceptual framework for participatory stories by defining story (narrative) and fiction in cognitive terms.
These were then used as building blocks for participatory stories,
also defined with cognition as the central component: choice. Having formulated the distinction between participatory stories and
non-participatory stories in cognitive terms, the next step is to look
more closely at these differences. Part I I of this book deals with
empirical studies of audience cognition.
Part II
Empirical studies
5
CHAPTER 5
A method for generating data for
exploring off-line cognition
T plorative method to empirically study cognition in relation to
HE PURPOSE
of this chapter is to present and discuss an ex-
participatory stories. As an instance of participatory studies, a work
of interactive fiction was used (see Chapter 4 for a classification of
participatory stories). The reasons for this choice are accounted for
in the method section below, but in summary it can be said that
interactive fiction is a prototypical example of participatory stories.
The study was designed to provide rich data for off-line cognition,
that is, the results of previously having experienced a participatory
story.18 The data consisted of speech and gesture when participants
talked about a participatory story (and other sources as comparison)
after having experienced it. To some extent, logged interaction with
the participatory story in the form of transcripts was used. The
analyses themselves are present in subsequent chapters: spatial
memory (Chapter 6), memory qualities (Chapter 7), and perspective
(Chapter 8). In this chapter, what is common to the method of
these analyses is accounted for: selecting, collecting, transcribing,
and coding data are described, motivated, and discussed.
Before the details of the method are presented, the natural context of using interactive fiction will be discussed.
Off-line is contrasted with on-line and is used in the sense it has in research on reading comprehension, see, e.g., Trabasso and Suh (1993) and van den Broek, Lorch,
Linderholm, and Gustafson (2001).
18
93
94
Participating in a story
5.1
A CONTEXT ANALYSIS OF USING
INTERACTIVE FICTION
Before beginning any laboratory study, particularly ones dealing
with people, it is wise to ask the following question: What does the
phenomenon that one is trying to study look like when it occurs
naturally? This is a question that needs an answer before one starts
to constrain the problem, remove influencing factors, and make the
situation into an artificial one more controlled and suitable for
study. This is important to consider because one does not want to
distort the situation too much and make it too artificial, and
thereby perhaps change the phenomenon itself. Methodological
considerations are crucially determined by the nature of the phenomenon.
There has been criticism of methods where certain variables of
interest are selected and studied in isolation, usually referred to as
laboratory studies. The argument is that in order to study some
factors one has to exclude other factors that might have an influence
if they were allowed. This exclusion would make the result applicable only to the constrained study situation, and not to normal, everyday situations (e.g., Guba & Lincoln, 1994). However, this need
not be a problem with laboratory studies per se, only with careless
application of them. The problem can be remedied by being careful
when making the transfer from natural to laboratory setting, or
‘context stripping’ as Guba and Lincoln (1994) call the process. For
instance, in research on narrative comprehension, it has been demonstrated that the type of task affects the extent to which people
construct mental models of the narrative (Wilson, Rinck, McNamara, Bower, & Morrow, 1993). So it is important to carefully consider what kind of instructions is given to the participants in an experiment.
What is the typical situation when a person is using a work of interactive fiction? Which factors are important and which are not? In
this section some important influencing factors of the phenomenon
to be studied will be identified—as it typically occurs. These factors
constitute the context and will be divided into the subgroups physical
A method for generating data
95
context, temporal context, personal and social context, and stimulus
context. The issues of this discussion come from informal sources:
talks with people who have experience with interactive fiction as
well as from the author’s personal background. Despite their informal sources, these points can generate important and fruitful
discussion of context stripping in the case of interactive fiction.
In the following discussion, the term ‘session’ is used to refer to
the uninterrupted period in which a user interacts with an interactive fiction system.
5.1.1 Contexts
5.1.1.1 Physical context
The immediate physical context is constituted by the user sitting in
front of a personal computer, running the interactive fiction system.
The phenomenon is likely to occur in a room in a home environment where the overall feeling is probably one of relaxation and
very low stress, although occasional disturbances such as phone calls
and nearby people may cause interruption. The user who normally
uses interactive fiction probably has a positive attitude toward the
computer as well as sufficient knowledge of how to operate it. If the
user is running several programs at once on the computer, there
might be an element of split attention. For example, if the user is
browsing files on the Internet at the same time as using interactive
fiction, there are likely to be effects on attention.
As the content of the work of interactive fiction can be quite
complex, the user might need to take notes. This could be, for instance, drawing a map of the landscape, writing down secret codes
or what character was where at what time. Without the possibility
of memory aids, the user might ‘get stuck’ in a certain position in
the work, or might wander about aimlessly.
Since interactive fiction usually is in English, persons with Swedish as their first language may look up unknown English words in a
dictionary.
96
Participating in a story
5.1.1.2 Temporal context
The user selects which work to use. Information about the work
may be obtained from mass media, the product packaging, and material included with the work (such as a manual). The user’s choice
leads to high intrinsic motivation. The user is expecting a pleasant
and exciting experience.
The typical consecutive period of use of a work of interactive
fiction is several hours. It is used in this way during a period of several days or weeks. Longer pauses (months or years) might also
happen, after which the work is used again. There is transfer of
knowledge between sessions, perhaps much the same as when a
reader of a traditional book picks up the book again after some time
and continues to read.
5.1.1.3 Personal and social context
It is proposed that the typical user of interactive fiction is characterised by being a young adult or adult, having Western cultural background, being literate, experienced with computers, literature, films,
and interactive fiction.
Why do people use interactive fiction? Even though there could
be some educational applications of interactive fiction (such as learning a second language), the main reason is that people enjoy the experience. People use interactive fiction for reasons similar to those
why people read novels for pleasure.
It is assumed that it is most common for a user to be alone while
using a work of interactive fiction, even if situations of two or more
people also occur. In earlier empirical research on computer games,
often two or more persons have been playing the game (e.g., Johansson, 2000; Linderoth, 2004). There could be at least two reasons
for this. First, it is a convenient way of obtaining data by recording
the conversation between the participants. Second, the choice of
more than one participant at a time may have been made because
the research questions concerned more than one participant, such as
how children can learn collaboratively by playing computer games
together.
A method for generating data
97
In situations of two or more people, collaboration on how to
make progress in the work of interactive fiction would be a common element.
In a single-user situation, and possibly also in multiple-user situations, the user is free to perform actions without the usual social
moral rules, which may influence the interaction by the use of socially inappropriate actions in relation to the work of interactive
fiction (such as trying to kill, abuse, or perform sexual acts on the
characters). These actions are likely to be absent in a study situation
where the user is aware that her actions are being recorded.
5.1.1.4 Stimulus context
An authentic work of interactive fiction is complex; it is authored
to captivate the user and be interesting and enjoying, both in content and in form. This makes it very doubtful to abstract any part
or feature of such a system to construct artificial stimulus material.
An authentic system constitutes an important factor in a study.
Which interactive fiction work is used in a study is important to
consider because works are associated with values. For instance, a
work where the user plays, say, a murderer rather than, say, an antinuclear activist could be considered negative by some people. The
number of existing interactive fiction works can be estimated to
range between a few hundred and a thousand.19 The majority of
these would however probably be judged to be of poor quality by
most users of interactive fiction. By choosing a work that has
proved to be popular, user motivation should be higher. The study
results can also be of more interest to the group of people who
normally use this kind of works (this is what Guba & Lincoln,
1994, call ‘the etic/emic dilemma’ of research).
No prior attempts to estimate the number of interactive fiction works in existence
were found. To arrive at this informal estimate, the World Wide Web was browsed
for information about commercial and non-commercial works, starting at the Interactive fiction archive, http://www.ifarchive.org.
19
98
Participating in a story
5.1.2 Moving from natural setting to study setting
Which factors are most affected when the natural context is removed, and what influence will this have on a study? Most important are those factors that might alter the phenomenon in a study
situation.
Concerning the physical context, the most important factor that
is considered to influence the study is the fact that the user is no
longer in a familiar home environment, but in a laboratory. This
might make the user nervous, and at least make the enjoyment less
than what is possible in a home environment. The solution was to
keep the study environment as close to a home environment as possible.
Another major difference between the natural situation and the
study situation is that the user is no longer permitted to choose
what work she will experience (and possibly also at an inconvenient
time). One possibility is to let the user (at a time convenient for the
user) choose the work, among a few alternatives. This of course
leads to methodological problems when comparing the results from
users who have chosen different works, which makes comparison
difficult. The solution was to use a single work of interactive fiction,
but to carefully choose an interesting work, thereby maintaining a
high level of user interest.
A final factor, which perhaps is the most influential one of all, is
that the user is being observed by a researcher. The researcher could
leave the room, but the user’s awareness of the study context can be
distracting and lead to unnatural behaviour. Also, for any study of
on-line cognition (such as what is taking place while using the work
of interactive fiction), some recording needs to be made (such as
computer activity logging or audiovisual recording). To the extent
possible, unobtrusive methods of recording should be used. For
studies of off-line cognition (the results of having used a work of
interactive fiction), recording is not necessary, and distraction from
recording is not an issue (unless the user suspects her behaviour is
being secretly recorded).
A method for generating data
99
The conclusion of the context analysis for the purpose of study
design is that the study situation should be as natural as possible and
the task should be to use a work of interactive fiction in a manner
normally done. The study should include an actual work of interactive fiction, even if this leads to a more methodologically difficult
design.
5.2
LANGUAGE AS DATA FOR OFF-LINE
COGNITION
Studying off-line cognition means looking at what happens after the
process of comprehension has taken place. It is the study of the results of comprehension, for example, mental representations in
long-term memory of people who have experienced a participatory
story. This is contrasted with on-line aspects which are dynamic and
take place at the time of comprehension, such as attention, working
memory and the dynamic activation and deactivation of inferences.
One way to study thought is through language. What a person
says and how she says it can show us much about what is going on
in that person’s mind. Speech can tell us about the content of
memories, accuracy, as well as whether certain kinds of details are
remembered better than others. The choice of lexical items can reveal conceptualisations. The choice of pronouns can reveal what
perspective a person adopts on some event described. Besides
speech, the spontaneous gestural activity that routinely accompanies
language production is highly systematic and gives insights into a
person’s cognitive processes (McNeill, 1992).20 Note that the reference to gesture here is not to typically symbolic gestures, such as
making a ‘V’ sign with your fingers to signal victory, but rather
gestures that are generated mostly in an automatic, not conscious
manner during speech. For instance, speech can show what people
consider important in an episode they are talking about, and gesThis holds regardless of whether one assumes, like McNeill, a single underlying
speech-gesture system, or separate systems for speech and gesture.
20
100 Participating in a story
tures can show how they mentally spatially locate themselves in
relation to things being described; as an observer looking at the
scene from the outside, or within the scene, as a character performing some action. Gesture is particularly interesting as a window to
cognition because it is mostly non-conscious, thus minimising conscious factors such as elaboration and inhibition. Conscious elaboration of memories can lead to a result that is not so much about the
original memory, but a product of conscious inferences, guesses,
and expectations. Inhibition can act to suppress overt behaviour
which would otherwise serve as evidence of some cognitive process.
The study of gesture enables a view of cognition that is in this way
more primary.
These are the reasons for carrying out a study of language when
people talk about remembered events. It is a fruitful way to learn
about many aspects of off-line cognition. Particularly, we can learn
what constitutes cognition involved in thinking about participatory
stories, and how it may differ from cases with non-participatory
stories.
In order to obtain speech and gesture as a basis for study, informal interviews were carried out with eight persons. The participants
were asked about events from five different sources which they had
experienced earlier (details are given in the methods section below).
The goal was to make the interview situation as close as possible to
an authentic conversational situation. Participants would talk as
freely and naturally as possible, in order to provide a rich base for
analysis of conceptualisations and memories. Audiovisual recordings of participants’ verbal and non-verbal behaviour were
made and transcribed. This allowed careful analysis of speech and
gesture. The data were analysed from a variety of viewpoints, which
is described in chapters 6 to 8. The nature of participants’ spatial
memory of a participatory story is the focus of Chapter 6. In Chapter 7, the phenomenal qualities of memories from participatory
studies and other sources are compared, as they surface in the verbal
accounts of events given by the participants. Finally, in Chapter 8,
the perspectives that participants adopt in relation to the events and
actions performed in a participatory story are investigated. Who
A method for generating data 101
carries out actions—the participant herself, the participant’s agent in
the participatory story, or a character viewed in a third-person perspective? Here, expressions in speech and gesture are studied as indications of perspective.
One has to keep in mind that the data about memory generated
by this method of investigating language may differ from how
memory is manifested in an authentic situation. In this book the
parts of memory that are visible through language are studied. Although this can reveal both conscious and non-conscious parts of
memory, it does not reveal how memory is used in a natural situation. There may be parts of memory that are behaviour-based, such
as procedural and implicit memory, which would only show in a
situation that includes action. However, this is a methodological
trade-off that was selected in favour of language as data on off-line
cognition.
The following section presents and discusses the empirical
method used to collect data for the analyses carried out in this
book.
5.3
METHOD
5.3.1 Material
5.3.1.1 Overview
This study starts with the assumption that a fruitful way to investigate how people remember and think about participatory stories is
to study speech and gesture when people talk about participatory
stories. In order to know what is particular about talk about participatory stories, these are contrasted with non-participatory stories. As a case of comparison, a traditional short story was used in
the study. But since there are some inevitable differences between
any given short story and any given participatory story, besides the
feature of participation—such as the subject matter and style—a special non-participatory version of the work of interactive fiction was
102 Participating in a story
also prepared (and this appeared similar to a short story). The types
of stories discussed so far could be said to be fictional, so a type of
narrative that is non-fictional is useful as a third contrast. A personally experienced episode is an example of this third kind. However,
these events could have happened a long time ago and could have
been retold and reshaped in memory many times. Thus, an additional set of laboratory tasks was used as a way of gaining more control over the events experienced. In summary, the source conditions
used were the following:
•
Game: Events from a computer game which the participants played.
•
Story: Events from a short story which the participants
read.
•
Personal: Events that took place earlier in the participant’s
life.
•
Tasks: Events from three tasks carried out in a laboratory
setting.
•
Non-participatory: Events from a special, adapted nonparticipatory version of the computer game, printed on
sheets of paper, which the participants read.
Table 5.1 shows a summary of the factors of the five source conditions used in the study. The possible impact on memory of the five
source conditions is discussed in Section 5.5.1.
Early research on reading often used short experimenter-created
texts as stimuli. These texts were highly artificial and lacked many
aspects of texts found in naturally occurring situations, such as being aesthetically enjoyable, interesting, and motivating. It has been
widely recognised within the field of reading research and discourse
processing that research should rather include authentic texts, such
as published novels and short stories, even if this means a decrease
in control of influencing factors (Graesser, Swamer, & Hu, 1997).
A method for generating data 103
Consequently, for the collection of language data here, a real interactive fiction computer game and a real short story were used.
Table 5.1. Summary of factors of the five conditions in the study.
Source condition factor
Source
condition
Fictive or
not
Participatory
or not
Approx.
time spent
Elapsed time since
encoding in memory
Game
fictive
participatory
1–2 hours
recent (0–2 hours)
Story
fictive
not particip.
20 min
recent (0–2 hours)
Personal
non-fictive
participatory
unknown
1 day to several years
Tasks
non-fictive
participatory
30 min
recent (20 min)
Nonparticip.
fictive
not particip.
20 min
recent (5 min)
Note: Fictive refers to whether the events should be evaluated in relation to the real
world or not (see Chapter 3), Participatory means that the participant had the ability
to make choices in relation to event sequences (see Chapter 4), Time spent means how
long the encoding situation lasts, Elapsed time since encoding gives the time between
encoding (i.e., experiencing or reading about the events) and retrieval (i.e., talking
about the events).
The general requirements for the stimuli for the five types of
sources are that they should be as similar as possible regarding content, complexity, elapsed time since encoding in memory, degree of
elaboration, and rehearsal. However, all these requirements could
not, for various reasons, be fulfilled. Also, given the explorative
nature of this study, it was not crucial that all of them were fulfilled.
Details are given below.
5.3.1.2 Interactive fiction
A work of interactive fiction entitled Anchorhead (Gentry, 1998)
was used in the study (in the following, this title will be used to refer to the work of interactive fiction used in the study). Anchorhead
simulates a world in which the player can control the main character to move around, and interact with objects and other characters.
It can be classified as typical piece of interactive fiction (sometimes
104 Participating in a story
referred to as an ‘adventure game’), see, for instance, Aarseth (1997).
It is a single-player game and does not connect to any computer
network. Figure 5.1 shows the location of Anchorhead in the general
classification of participatory stories from Chapter 4.
Anchorhead consists entirely of English language text. It outputs
descriptions of locations, objects, persons, and events, and the
player types commands using close-to-natural English sentences on
the computer keyboard. Anchorhead contains places to explore,
people to interact with, objects to manipulate, and an underlying
plot that drives the action and sets goals for the player. See the appendix for a transcript of interaction with Anchorhead.
participatory stories
pretence
non-participatory
stories
live role-playing games
Aarseth’s ergodic literature
Aarseth’s
cybertext
fairytales
novels
computer games
short stories
interactive fiction
Red circle
■
adventure games
Anchorhead
■
plays
cinema
board role-playing games
everyday narrative
Aarseth’s & others’ hypertext
Figure 5.1. Location of the work of interactive fiction, Anchorhead, and the short
story, Red circle, in the general scheme of participatory stories as defined in Chapter
4.
A method for generating data 105
Anchorhead ran on a standard personal computer (Intel Pentium I I I
C P U , Microsoft Windows 98, 17 inch screen, and standard P C keyboard) using the WinFrotz Z -machine interpreter software (Lawrence, 1999). The text size was set to larger than the usual text size
on the system (such as for menus and window titles) in order to be
clear and easily readable. The colour of the background and the text
could be set to the personal preferences of the participant. Figure
5.2 shows a screen dump from the WinFrotz software running Anchorhead.
Figure 5.2. A screen dump from the WinFrotz software running Anchorhead.
106 Participating in a story
When the work of interactive fiction was selected for the study,
these were the main requirements:
•
The work of interactive fiction should have a strong narrative element and plot, in order to be comparable to a short
story.
•
It was desirable that the work should be considered good
within its genre. The work used was considered excellent
by a vote of the interactive fiction community.21
•
It should be recent, because more recent works of this type
are technically better than old ones at handling the user input, thereby making the interaction smoother and the
work more enjoyable. It is also more likely that participants have not already been exposed to a new work than an
old one.
•
It should be text only, and not contain images, audio, or
video, to be similar to the short story and make them easier
to compare. Any differences found would be easier to attribute to content because both interactive fiction and short
story are text only.
•
The content of the work should be realistic and not too
fantastic, so that the events could be compared to people’s
own personal experiences.
•
It should include at least one other character, besides the
main character, in order to compare with short stories,
which typically contain several characters.
•
Participants should be able to finish a relatively autonomous part (comparable to a chapter in a book) of the work
within 1–2 hours (longer time could tire the participants
and would make the study unmanageable).
In the Xyzzy Awards of 1998, winner of Best Setting, and finalist in the following
categories: Best Individual N P C , Best Puzzles, Best Writing, Best Game, Best Story
(Mullin, 1999a,b).
21
A method for generating data 107
•
Participants should not have experienced the work before.
•
The contents should be neutral in that it should not contain arguably charged matters such as politics, gender issues, eroticism, etc., which could offend participants.
•
The language used should be easy to read and understand,
so as to maximise memory encoding, and also make the
participant able to make progress in the work.
Anchorhead was seen to fulfil most of these requirements; however,
some aspects of this work were less than optimal:
•
It is somewhat slow and the plot does not advance by itself
(there are other works where the plot advances by itself,
even if the player does not perform actions to advance the
plot).
•
It had many locations to visit—this could be confusing and
difficult to remember.
•
The main character is female, so the player controls a female character. Could that make it difficult when the participant is male?
However, the interview revealed that all participants enjoyed playing Anchorhead and thought it was exciting. They did not have
much difficulty finding their way around and (the male participants)
were not bothered by the fact that the main character was a woman.
No other material, such as dictionaries or paper and pencil, was
provided in order to reduce the complexity of the study. Possible
consequences of this are discussed below in Section 5.5.3.
One artificial constraint was that only the first part of the Anchorhead was used in the study, because of time limitations (the full
work typically takes dozens of hours to complete).
It is difficult to simply state the length of Anchorhead, say, in a
measure such as the number of words. The length of Anchorhead is
relative to how it is used. To get a rough estimate of the textual
108 Participating in a story
length, one can look at the length of the log files that were generated while participants interacted with Anchorhead, which shows all
textual activity from Anchorhead as well as from the participant,
although this includes a large portion of repeated text output by the
game. Another measure of the length of Anchorhead is the amount
of time spent using it. Time spent on Anchorhead and length of the
resulting log file are presented below in the section ‘Overview of
data’ (Table 5.5). Thus, regardless of the measure used, the length of
Anchorhead varies among participants. This is an inevitable consequence of what participatory stories are (see Chapter 4).
5.3.1.3 Short story
The short story used in the study was the first part of a Sherlock
Holmes tale written by Arthur Conan Doyle (1911/2000) called The
adventure of the red circle (in the following referred to as Red circle).
It consisted of 5,702 words and was printed in 15-point Times New
Roman typeface on 14 white sheets of A4-sized paper, stapled together. The title of the short story was presented at the top of the
first page. The complete original short story consisted of 7,409
words. It was judged to be unnecessarily long for the study and was
cut off at a place that was judged to be a natural end-point. In that
respect, it had a parallel with the computer game, of which again
only the first part was used.
In Figure 5.1, the location of Red circle can be seen in the general
classification of participatory stories from Chapter 4.
These were the main requirements when selecting the short
story:
•
It should be text only (no pictures), to be similar to the
game.
•
It should include at least two characters.
•
It should tell about events in grammatical past tense (or
perhaps present tense)—future tense makes it too different
from the other source conditions.
A method for generating data 109
•
It should not have very unusual or bizarre contents, because it might make it stand out in memory too much in
comparison to the other two conditions.
•
It should include some events that could make the participants use narrative spatial gestures (because this was an area
that was planned to be explored in the analyses later), for
instance, someone going somewhere, spatial relationships
between objects, and positions of characters that are important to the story.
•
It should be in English, because the game was in English.
•
It should be easy to read and understand.
•
It should take about 15–30 minutes to read in order not to
make the total study session too long and tiring for the participants.
•
It should not be well-known; participants should not have
read it or heard about it before.
•
It should not contain arguably charged matters such as politics, gender issues, eroticism, etc., that could offend participants.
Many stories are available both as a printed text and as a computer
game, such as J. R. R. Tolkien’s The lord of the rings and Douglas
Adam’s The hitchhiker’s guide to the galaxy, which seems to present
ideal cases of stimuli on which to base a comparison. However,
these were considered unsuitable because they are well known and
are likely to have been read or played by the participants. The uncontrolled elaborations and reading and talking about these stories
would introduce a major bias in the study. The chosen work, Red
circle, was similar to Anchorhead in that both contained problems
and puzzles to be worked out.
110 Participating in a story
5.3.1.4 Non-participatory version of the work of interactive fiction
Because of differences between Anchorhead and Red circle, besides
participation, such as contents, style, description of spatial arrangements, etc., a special, non-participatory version of Anchorhead was
prepared. In this way, it was possible to make a closer comparison
of how participants talked about spatiality and character perspective
as influenced by the factor of participation. The text output of a
standard session with Anchorhead was captured (it was compared to
actual sessions from participants 1–4). Some minor changes were
made to the text in order to make it readable as a short story:
•
Repeated descriptions of environments (a standard feature
of the work of interactive fiction) were removed.
•
Some minor stylistic modifications were made, such as
summarising some repetitions and joining some passages using small phrases.
At the top of the first page was written ‘Anchorhead’ followed by
‘A gothic by Michael S. Gentry’. The text was 5,885 words long,
printed in 16-point Times New Roman typeface on 19 white sheets
of A4-sized paper, stapled together.
5.3.1.5 Laboratory tasks
In order to elicit talk about real, personally experienced events,
questions were put in the interview about things that had happened
to the participants. However, personally experienced events may
have been thought about and retold before, and therefore reshaped
in memory (Labov & Waletzky, 1967; Schank, 1990; Tversky &
Marsh, 2000; Niedźwieńska, 2003). It is also the case that events
which took place more than a day or two ago differ in how they are
represented neurologically compared to very recent memories
(Conway, 2002). These two reasons made it unsuitable to rely on
personal experience as the sole case of comparison. To balance this
situation, a number of laboratory tasks were devised. These tasks
were designed to produce personal experience in a more controlled
manner.
A method for generating data 111
The requirements for selecting tasks were the following:
•
Tasks should include as many sensory modalities as possible and not just be paper-and-pencil tasks, in order to be
comparable to prior personal experience.
•
Tasks should not be fictional in nature or be experienced
through any media such as text, images or video, in order
to ensure self-experience.
•
Tasks should involve a variety of skills: visual, mathematical, artistic, etc., although no expert skills.
•
Tasks should be motivating and fun.
•
Tasks should be manageable, that is, not take too long, and
be possible to carry out inside one room.
In order to increase motivation and make it possible for participants
to perform the tasks that they have an interest in, participants were
allowed to select three out of five tasks to carry out. A large number
of tasks were pilot-tested, and reduced to a list of five that worked
practically:
•
Determine the size of the room, given a paper, pencil, a
piece of string, and a ruler.
•
Arrange five small metal objects from lightest to heaviest.
•
Guess the contents of five sealed opaque plastic jars (containing sugar, rice, etc.).
•
Tie three types of knots on three pieces of rope from pictorial instructions.
•
Make an abstract collage out of magazines using glue, paper, and scissors.
5.3.1.6 Interview questions
The idea of the interview questions was to bring out as much as
possible of the participants’ memories, for instance, what they re-
112 Participating in a story
membered about events, how they viewed events, and about the
spatial details of the events. This was done by guiding them as little
as possible and allowing them to speak freely.
The following criteria were used for selecting interview questions. The questions should
•
elicit talk by the participant which includes gesture, especially spatial gestures
•
make the participant want to tell about events in the five
conditions
•
be informal and make the participant as relaxed as possible
•
focus on content (not meta-work features, i.e., how the
game was programmed or how the short story was written)
•
not reveal or help participants guess the actual research
questions
•
not influence the participants with regard to the research
questions—questions should not contain pronouns or perspective (e.g., ‘Did you…’ or ‘Did the character…’).
The interview questions were formulated to be as neutral as possible. The participants were asked ‘to tell what happened in the game’
and ‘what happened in the short story’. The interview questions
about the participants’ personal experience were linked to events
from the computer game or the short story, asking whether a similar event had happened to them personally. This way, the conversation moved on naturally, and material suitable for comparison was
obtained.
5.3.1.7 Study locations
The study was carried out in rooms which all were arranged to be
neutral and home-like, so that the participants would feel as relaxed
as possible, in order not to distort natural use of interactive fiction,
reading, or conversational behaviour.
A method for generating data 113
For the participants who played Anchorhead, the initial briefing
and using Anchorhead took place in a laboratory with a personal
computer. Participants had the option to close the door to be alone
while using Anchorhead. Lights were dim and it was ensured that
they did not reflect in the computer monitor. Reading of the short
story took place in a room with normal lighting. Participants had
the option of closing the door to be alone while reading.
The participants who carried out laboratory tasks had the
briefing, the tasks, and the reading in the same room. The room had
normal lighting, closed door and a window to a control room.
The interview was conducted in a room usually used for video
recordings. Two soft chairs were arranged facing each other on opposite sides of a small coffee table, where glasses of water were
available. The room had normal lighting conditions.
All rooms had normal indoor temperature and ventilation.
5.3.1.8 Recording equipment
The participant’s actions in the computer game were logged on the
computer as text files with a built-in function in the WinFrotz Z machine interpreter software (Lawrence, 1999).
For recording the interviews, a mini-D V digital video camera
with built-in stereo microphone, placed on a stand, was used.
5.3.2 Participants
Eight persons, four males and four females, aged between 20 and 36
participated for pay. The participants were students recruited at
Göteborg University, Sweden.
The general requirements for selecting the participants were the
following:
•
They should be adults. Cognitive functioning should be
fully developed and they should have normal Western cultural background knowledge.
•
They should have Swedish as their first language.
•
They should have reading and writing skills in English.
114 Participating in a story
•
They should not have any physical disability that stops
them from moving their arms and upper body.
•
They should not have been exposed to the work of interactive fiction or the short story prior to the study.
For the first four participants, who used Anchorhead, the additional
requirement was that they should have some experience with other
works of interactive fiction similar to the one used in the study, so
that they would not need to be trained before participating in the
study (this requirement was mentioned during recruitment). All of
the above requirements were checked, either at recruitment,
through the main interview, or through a questionnaire. Table 5.2
summarises the variables of the participants in the study.
A design of two groups of four participants each was used because the same eight participants could not participate under all
conditions. This would reveal the underlying study idea and make
the situation unnatural, and in some cases confound the effects of
the conditions (e.g., having participants play the game and read the
non-participatory version of the game).
Most participants had Swedish as their first language, and the remaining two who did not were judged to have an excellent knowledge of Swedish as determined from the interviews. The English
proficiency of the participants was good or excellent (as judged from
the interview). (If a participant commented about at least one problem in understanding the English in the game or the short story,
they were rated ‘good’ and if they did not report any problems with
the English, they were rated ‘excellent’.) A further indication of
sufficient background knowledge of both English and works of interactive fiction was the fact that all four participants were able to
successfully finish the required part of Anchorhead.
It turned out that participant 3 had played the computer game
prior to the study. However, his data were treated in the same way
as the others’ data. Discussion of and arguments for this may be
found in Section 5.5.4 below.
A method for generating data 115
Table 5.2. Participants’ variables.
Participant variables
Participant
Gender
First language
English
proficiency
Unknowing
interviewer
IF experience
1
male
Polish
good
no
a few works
2
female
Swedish
excellent
yes
many works
3
male
Swedish
good
yes
many works
4
male
Swedish
good
yes
a few works
Participant
Gender
First
language
English
proficien.
5
female
Russian
good
yes
1/week
medium
6
female
Swedish
excellent
yes
>1/week
medium
7
female
Swedish
good
yes
>1/week
medium
8
male
Swedish
excellent
yes
>1/week
medium
Participant variables
Unknowing Reads
interviewer fiction
Difficulty
of text
Note: Unknowing interviewer refers to whether the participants believed that the
interviewer did not know about the game or story they were asked about (this was a
consequence of the study design). IF experience refers to how extensive experience of
interactive fiction the participant had. Difficulty of text refers to how the participants
rated the difficulty level of the text. Reads fiction refers to how often the participants
read fiction. These two last factors were self-rated on a questionnaire.
116 Participating in a story
SOMETIME PRIOR
TO THE STUDY
PARTICIPANTS
1–4
0. REAL
EVENTS
PART OF STUDY
1. PLAY
GAME
2. READ
STORY
3. INTERVIEW
PARTICIPANTS
5–8
1. LAB
TASKS
2. READ
’GAME’
Figure 5.3. Overview of the study procedure.
5.3.3 Procedure
The procedure differed for the first four participants and the last
four participants. Figure 5.3 shows an overview of the procedure.
The first four participants, one at a time, were taken through the
following steps of the study:
1. The participant was given a brief introduction to the study. The
participant was told she were going to use (the first part of) a
work of interactive fiction, then read a short story, followed by
a talk where she was going to be asked about her opinions on
the work of interactive fiction and the short story. She was told
that the experimenter did not know the contents of the work or
the story.22 The words memory, gesture, spatiality, or perspective
were not mentioned during the study in order not to influence
the participants.
2. The participant was instructed to use the work of interactive
fiction as she would if she were doing it alone, at home. The participant began and the starting time was noted. Every 15 minutes, a check was made by the experimenter on the participant’s
For participant 1, the experimenter was not told to be unknowing of the contents
(due to experimenter error).
22
A method for generating data 117
progress. Hints, in the form of keywords, one by one from a
list, were given if needed to help the participant make progress.
This elaborate setup was used so that the participant would still
believe that the experimenter did not know the contents. When
the participant reached the end of the first part of Anchorhead,
the time was noted. The amount of time each participant spent
with Anchorhead is presented below in the section ‘Overview of
data’. All participants were able to finish the required part of
Anchorhead.
3. The participant was taken to another room and given a short
story to read silently. Starting and finishing times were noted.
4. The participant was informed that the main part of the study
was over and that she now was going to be asked about her
opinions on the work of interactive fiction and the short story.
A semi-structured interview followed in another room. Questions guided the interview, but if a new topic was introduced by
the participant, there was room to pursue it. A video camera,
filming the participant, placed at an angle from the participant
not directly in front of her or him (in accordance with McNeill,
1992), was turned on. The participant was told that the video
camera was used for convenience, instead of the interviewer taking written notes. The participant was not told that the interview would be transcribed. The interview was conducted in a
conversation-like manner. The order of questions was fixed, but
some improvisation was done to move between or skip questions, in order for the conversation to be more natural (see section ‘Interview questions’ above). The interview questions were
arranged so that talk about events from the five conditions
would be mixed. This was to avoid the situation where the participant would become more relaxed as the interview proceeded
and perhaps talk more and gesture more frequently, which otherwise could introduce a systematic influence on how the participant talked about the conditions.
118 Participating in a story
5. After the interview, participants received pay and were fully
debriefed. Participants were told not to discuss the study with
other people.
For the remaining four participants (nos. 5–8), the following steps
were carried out for each participant in turn:
1. The participant was welcomed by an assistant and seated in a
separate room which had a window to where the assistant was
sitting. A paper with written instructions introduced the study
and requested the participant to select three out of five available
tasks, by looking at a descriptive keyword of each task.
2. The assistant provided the participant with four boxes, three
containing the tasks and the fourth containing the nonparticipatory version of the work of interactive fiction.
3. The tasks were dealt with in order, the non-participatory story
coming last.
Steps 4 and 5 were the same as for the first four participants. The
exception was that the interview was carried out by a new person,
not present in the first part of the study. The arrangement of an
unknowing interviewer was chosen in order to make the interview
situation closer to a real conversational situation. The participant
would find it more natural to talk about what happened when he or
she believed that the interviewer did not know about it.
The procedure ensured that a sufficient amount of time passed
between encoding and retrieval. Information about the game or the
story was lost from short-term memory, and could only be accessed
from long-term memory at the time of the interview.
The ordering of first Anchorhead, then reading the story (for the
first four participants) and first carrying out tasks, then reading the
non-participatory version of Anchorhead (for the last four participants) was fixed for all participants. This ordering was used because
it removes any risk that the participants might think that clues from
the short story should be used while solving problems in An-
A method for generating data 119
chorhead, or that clues from the non-participatory version should be
used for solving the laboratory tasks.
The entire procedure was carried out in one sequence, rather
than being split up across two or more days. This minimises several
risks: First, rehearsal leading to biased strengthening of certain parts
of memory is minimised. Second, there is no risk that the participant finds Anchorhead on the Internet and plays it in an uncontrolled manner. Last, the participant is stopped from discussing Anchorhead with other people. There are at least two risks with the
procedure used in the study (without pauses): fatigue, affecting attention and memory encoding, and interference, leading to confusion between Anchorhead and the short story. However, no evidence of these two drawbacks was seen in the data.
An important design choice was to instruct participants that they
were going to talk about how they liked the game and the short
story. This makes it unlikely that they anticipated a memory test,
and the playing and reading could take place in a way close to an
authentic situation.
5.3.4 Transcription and coding
The transcription of speech and coding of source condition segments are presented here. More detailed coding of speech for various analyses is described in Chapters 6–8, together with the relevant
analysis. The transcription and coding of gesture are described in
Chapter 8.
5.3.4.1 Transcription of speech
Speech and non-verbal communication of the interviewer and the
participants in the auditory channel, such as emphasis, some intonation, laughing, smacks, and inhalations were transcribed. Other
non-verbal communication was considered to lack theoretical relevance to the research questions and was therefore ignored. Speech
was transcribed in Swedish standard written orthography using an
adapted Conversation Analysis standard. See Table 5.3 for a transcription key.
120 Participating in a story
Table 5.3. Transcription key for speech.
Element
Meaning
Description
(.)
pause
If a clearly perceptible pause is present. Pauses are only
transcribed as present or not, and no time is given.
(unclear)
uncertain
transcription
If speech is unclear, but still probably accurate, this is
placed within parentheses.
()
speech left out
If the speech is unintelligible.
under
emphasis
If a word or phrase is emphasised, it is underlined.
a[a]
[b]b
overlapping
speech
If speakers speak simultaneously, their speech is shown
left-justified in a column within brackets.
:
prolongation
Colons are used to indicate prolongation of the sound
just preceding them.
-
interruption
A hyphen directly after a word or part of a word indicate an interruption (often self-interruption)
!
exclamation
An exclamation mark at the end of a word indicates
that it is uttered louder than surrounding speech.
?
rising intonation
A question mark at the end of a word indicates a rising
intonation (not necessarily a question).
=
connected
speech
If the speech is connected (without pauses) with speech
on an earlier line, the two lines are ended and started
with an equal sign.
((a))
non-speech
Text within two parentheses indicates non-speech activity, e.g., ((smack)) for smacking with the lips, ((laugh))
for laughing.
ºh
inhalation
A ring followed by a number of h’s indicates audible
inhalation. The number of h’s shows the length.
5.3.4.2 Coding: Segmentation of discourse
The transcriptions were coded into five types of segments according
to which type of events or conditions they were referring to: Game,
Story, Personal, Tasks, and Non-participatory (see Table 5.1). Almost always, a segment was demarcated by the interviewer changing the topic with a question or a request, such as ‘now let’s talk
A method for generating data 121
about the short story’. The sum of the segment lengths for each
participant and condition is presented in the section ‘Overview of
data’ (Table 5.7).
5.4
OVERVIEW OF COLLECTED DATA
A general overview of the collected data is presented here (results of
separate analyses are presented in Chapters 6–8). First, times and log
files for Anchorhead are presented, then transcriptions and coding of
the interviews.
5.4.1 Times and log files for Anchorhead
Table 5.5 shows the time spent using Anchorhead for each participant, as well as the length of the resulting log file. An excerpt from
a log file created while using Anchorhead can be found in the appendix.
5.4.2 Interviews
5.4.2.1 Transcriptions
Table 5.6 shows the total and individual lengths of the transcriptions for the participants.
5.4.2.2 Coding
Table 5.7 shows the sum of the segment lengths for the five conditions and the eight participants. The segments for a source condition were rarely consecutive. As discussed, the methodological aim
was to mix conditions temporally—to the extent it was natural to
the conversational situation—in order to avoid ‘warm-up’ effects.
122 Participating in a story
Table 5.5. The time participants spent on Anchorhead and the length of the generated
log files.
Participant
Length (hour:min)
Length of log file (words)
1
2
3
4
2:16
1:04
0:22
1:24
14,910
26,104
10,288
26,518
Table 5.6. Length of transcribed material for the eight participants.
Participant
Transcribed length (min:sec)
1
2
3
4
5
6
7
8
TOTAL
23:14
15:10
16:57
32:10
13:50
6:04
8:12
7:28
123:08
Table 5.7. Sum of segment lengths of audiovisual recordings for the five source conditions and the eight participants.
Source condition
Participant
Game
Story
Personal
1
2
3
4
5
6
7
8
11:24
07:06
07:32
13:23
07:45
05:53
06:31
10:59
03:57
02:11
01:53
07:47
Note: Times shown in minutes:seconds.
Tasks
Nonparticipatory
04:34
01:03
02:43
01:40
09:16
05:01
05:29
05:48
A method for generating data 123
5.5
DISCUSSION OF METHODOLOGICAL ISSUES
In this section, various methodological issues will be discussed.
First, a detailed investigation is made of how memory is influenced
by the design of the study. Since memory is the central object of
study, it is crucial that no unwanted distortions introduce bias that
affects the issues under study. After that, ecological validity is discussed related to the naturalness of the playing, reading, and conversational situations. Further, a note is given on re-playing Anchorhead, and the possible effects it might have on the study. Finally, the issue of generalisability is addressed.
5.5.1 The implications of the study design for memory
Since the study taps into participants’ memories, it is of concern
how the study deals with various factors affecting their memory. At
what stages in the study is information encoded into memory, and
at what stages is it retrieved, and what possible pitfalls concerning
memory need to be guarded against?
The present study examines what happens after participants have
used Anchorhead, read the non-participatory version of Anchorhead,
read a short story, carried out laboratory tasks, and experienced
personal experiences. Within the participant’s cognitive system, it
can be said to be the last step in the chain, which starts with encoding in memory—in the playing or reading situation—and ends with
retrieval of information from memory—in the interview situation.
In this section follows a discussion of the information’s path from
start to finish, originating with stimuli outside the cognitive system,
which is encoded by the participant and ending with the data coding procedure (see Figure 5.4).23 This sequence is used in the following to structure the discussion of memory factors which could introduce bias in the study.
It is also reasonable to say that information flows in the other direction, from the
participant’s memory to the stimuli, as the participant carries out actions.
23
124 Participating in a story
Rehearsal
& narrative
construction
1.
Stimuli:
Anchorhead,
short story...
2.
Encoding
3.
Storage
4.
Retrieval
5.
Data collection,
transcription,
coding
Study situation:
interview
Figure 5.4. An overview of the flow of information in the present study. Information (1), enters the memory system (2–4), and is finally captured through recording,
transcription, and coding (5).
The point of this discussion is that either the design of the study
minimises any relevant memory distortions successfully, or the alterations of information play no significant role in relation to the
subsequent studies in Chapters 6–8.
5.5.1.1 Stimuli
Although much of the information is created in the interaction with
the participants, some portion of the information could be said to
exist outside the participants at this stage. The stimuli in the five
conditions have some notable differences. The computer game Anchorhead, the short story Red circle, and the non-participatory version of the computer game all consist of English text. The last two
of these are presented as printed text on paper, while the computer
game is text on a computer screen. The stimuli in the game condition also consist of a computer keyboard for typing text. The other
two conditions, the personally experienced events and the laboratory tasks, share the property that they occurred in the real world
and thus potentially involved the full perceptual repertoire, in terms
in sense modalities, of the participants.
A method for generating data 125
Nevertheless, the five conditions are similar in that they all can
be said to contain various actions in time (which in later cognitive
processing stages become represented as discrete events). ‘Action’
here means that someone is doing something. The conditions are
also similar in that they all consist of elements of these actions that
have various spatial relationships. A special spatial position, which
is present in all five conditions, is that of the experiencing self in
relation to other elements of the actions.
5.5.1.2 Encoding
The participant plays the computer game, or reads the short story.
Stimuli are encoded in memory through a constructive process, depending on the attention, goals, and background knowledge of the
participant. Another way of putting it is that meaning is created in
this stage. Stimuli not attended to are not encoded (save for subliminal perception, which is judged to be highly marginal in this
context). Purpose determines how attention is controlled, but also
how stimuli will be encoded (van den Broek, Lorch, Linderholm, &
Gustafson, 2001). In the current study, the purpose of reading or
playing for entertainment or interest was induced (as opposed to a
purpose of memorising details for a subsequent memory test). Background knowledge is what makes the participant able to understand
the stimuli, by relating it to earlier knowledge, for instance, using
memory schemas (Graesser, 1981), including expectations on narrative organisation (Bartlett, 1932). The participants had the required
background knowledge to understand the game, the short story,
and the laboratory tasks.
The time for encoding differed across the conditions. This is a
natural consequence of the nature of participatory stories, and a
consequence of letting participants read the short story at their own
pace. Since experiencing a participatory story depends on participation, this inevitably results in variation in encoding times among
participants. Again, this is nothing we should control, since it
would detract from the authenticity of the participatory story situation. Unless we design a study to incorporate a very high number of
participants so that we can match those with similar times, we are
126 Participating in a story
left with the variation in encoding times. The same is the case for
the short story and the non-participatory version of the computer
game. Reading should be at the participant’s own pace. Participants
were instructed to read through the texts once, thus minimising the
risk that repetition would be the cause of variation in encoding
times. Nor is it reasonable to control the time taken for participants
to carry out the laboratory tasks. The differences in encoding times
have consequences for what we can meaningfully study and for the
conclusions we can draw. Longer encoding times generally lead to
more successful retrieval, so comparing memory for details across
conditions would not be sound. On the other hand, if the natural
mode of experiencing participatory stories as compared to nonparticipatory stories entails different encoding times, and this in
turn leads to differences in retrieval, it would still be an empirical
finding worth showing.
As was discussed in Chapter 4, it is likely that the process of
event comprehension is similar across sense modalities, such as listening, reading, or watching, and also similar when comparing real
life events and events presented in a medium such as text.
5.5.1.3 Storage
Knowledge is held in long-term memory in the storage phase. Possible influences on memory at this stage are interference, rehearsal,
narrative construction, and fading. These will now be considered in
turn.
Interference from similar material encoded afterwards could occur so that material from the computer game interferes with material from the short story, for instance, mixing details up. Any of
these could also interfere with material already encoded by the participant at an earlier stage, such as other games played or other short
stories read. However, no evidence of such interference was found
in the collected data, which makes it plausible to assume that interference did not play a role in the present study.
Rehearsal can strengthen memory, but may do so differently so
that some parts of the information are emphasised more than others. Together with other background knowledge, rehearsal can also
A method for generating data 127
construct new or alter details from the original memory. One type
of rehearsal, narrative construction, could affect memory in major
ways (Schank, 1990). By thinking, using a narrative schema, about
something that was experienced, the material could be reshaped
according to features of the narrative schema. More prominently, if
the experience is talked about with other persons, it is reshaped and
repackaged through a constructive narrative process, not only
through production, but also through features of the conversation
as a whole, for instance, what other persons are saying. In this
study, the personal experiences are the type of material that is most
likely to have undergone these processes. For the other four conditions in the study, retelling experience to other people was not possible, and elaboration through thinking was kept at a minimum and
controlled by keeping the time between encoding and retrieval
short and roughly equal between conditions.
Fading could occur in the case of personal experience, but for the
other four conditions, it is unlikely to occur since the time between
encoding and retrieval is at most a couple of hours.
It should be stressed, however, since the veridicality of memory
is not central to the study, these processes are likely to play a small
role. It is argued that even if memory is altered and details are remembered incorrectly, the general manner of conceptualisation will
not be altered, at least not in a way that will affect the analyses carried out in Chapters 6–8.
5.5.1.4 Retrieval and language
During the retrieval phase, information is brought from memory
and the participant remembers. What memories are retrieved depends on the presence of cues (internal or external) as well as
physiological factors, such as the level of stress.
The study was designed to minimise stress. A high level of stress
might lead to poorer retrieval of details. Even if some participants
did feel uneasy, it is not likely to have affected their conceptualisations.
The framing of the interview questions could act as cues for specific memories. However, a biased retrieval of details is not likely
128 Participating in a story
to influence conceptualisations. The interview questions could affect
conceptualisation, but were carefully constructed not to (by not
mentioning pronouns, perspective, etc.).
The interview situation invokes a set of conversational codes,
which might affect retrieval in various ways. First of all, the purpose of the talk, that is, knowledge of the research questions, may
influence retrieval. This was not the case since the participants were
unaware of the purpose of the talk (as checked in the debriefing after the interview). Further, knowledge of who the participants were
talking to guided what information was recalled. How conceptualisations were verbalised by the participants may depend on their
beliefs about the knowledge of the interviewer. If they think the
interviewer is someone who knows a lot about computer games,
they might choose to express a different verbalisation than if they
think they are talking to a novice. Participants knew that the experimenter know about both computer games and short stories.
Importantly, the participants all believed that the interviewer did
not know the contents of the game, short story, laboratory tasks, or
non-participatory version, and this was the same across all conditions (with one exception, participant 1). This not only gives a more
natural conversation, but it is also likely to influence what context
and details are verbalised. But again, this does not arguably affect
the conceptualisations made by the participant.
As already discussed above, the present study investigates how
memory is retrieved for use in language. There may be differences
compared to how memory is retrieved for action, as would happen
in a naturally occurring situation when a person is using a work of
interactive fiction. The reason for choosing in favour of retrieval via
language is that it provides data that are interesting enough for a
study of off-line cognition.
5.5.1.5 Data collection procedures
The stages described above (2 to 4) all take place inside the cognitive
system of the participants (i.e., inside the head). The final step occurs outside and comprises the data collection procedure. Information is recorded, transcribed, and coded. Alterations of information
A method for generating data 129
in these stages is of course hoped to be minimal (discussions can be
found at separate sections on transcription and coding in this chapter).
5.5.1.6 Summary of the influence of the study on memory
To summarise, the five conditions can be considered to consist of
actions (someone is doing something), taking place in a spatial arrangement, which are interpreted by the participants through their
background knowledge and then memorised. Memory is later retrieved in the conversational situation as a result of the interviewer’s
questions. Although there are some differences across conditions,
such as time for encoding and retention time, as far as can be seen,
memory seems to be equally influenced in relation to the subsequent analyses carried out in Chapters 6–8.
5.5.2 Language and culture
There were several languages involved in the study, which may be a
complication. The participants were Swedish (or fluent in Swedish).
The instructions, given in speech as well as written in the case of the
laboratory tasks, as well as the interview, were in Swedish. This was
the most natural choice of language. The computer game was in
English because no works of interactive fiction that fulfilled the
study requirements were available in Swedish (English is the standard language for interactive fiction). The short story was in English
in order to match the computer game. By checking English language proficiency, sufficient comprehension of the computer game
and the short story were ensured. The switch from English to Swedish was not considered to be a problem for the participants.
Spatial gestures and spatial thinking are culturally determined
(Levinson, 2003). In the present study, the stimuli were in English,
but the interviews were carried out in Swedish. However, there
should be no problem, since both languages utilise the same linguistic frame of references (Levinson contrasts languages with absolute
linguistic frame of references with languages with a relative linguistic frame of references, such as European languages). The same
130 Participating in a story
holds for the cultural backgrounds of the participants. It turned out
that two participants had non-Swedish cultural backgrounds. However, these were considered to be sufficiently similar not to influence the study.
5.5.3 On the naturalness of the study
The study aimed to investigate playing the computer game and reading the short story, as well as talking about these, under naturalistic
conditions. Thus, it was important that the situation was as natural
as possible. The following discussion addresses the naturalness of
the playing situation, the reading situation, and the interview situation.
5.5.3.1 The naturalness of the computer game situation
Effort was expended to make the playing situation as authentic as
possible. First, the computer game was a real work of interactive
fiction. Second, participants were instructed to play as they normally do in order to say how they liked the game afterwards. Third,
the participants were experienced players of interactive fiction.
Fourth, the playing was carried out in a neutral room, so that the
participants would feel as relaxed as possible, in order not to distort
natural use of interactive fiction. Fifth, for the same reason, the participants were left alone in the room while playing. Finally, there
was no time pressure.
A few study design choices were made that may result in a situation different from an authentic playing situation. To begin with,
the participants were not allowed to choose which game to play.
This could have caused lowered motivation. However, as seen in
the interviews, all participants highly enjoyed playing the game.
Moreover, no other material, such as dictionaries or paper and pencil, were provided. This could have lowered comprehension and
memory somewhat, in that participants could not look up difficult
words in a dictionary. Neither could they support their memory by
drawing maps. However, again, as seen in the interviews, partici-
A method for generating data 131
pants seemed to have no problem understanding the language or
remembering the spatial layout.
5.5.3.2 The naturalness of the reading situation
Several design decisions were made in order to increase the naturalness of the reading situation (holding for both the short story and
the non-participatory version of Anchorhead). First, participants
were not instructed to memorise the text. They were simply asked
to read the text and afterwards discuss their opinion about it. Second, an authentic short story was used. Third, participants were left
alone while reading, so that they would feel more relaxed. Fourth,
participants could read at their own speed. Fifth, reading was done
from text printed on paper (and not presented on, e.g., a computer
screen). Finally, participants read the text once, since it was considered the most usual way to read a text in a real situation.
5.5.3.3 The naturalness of the interview situation
Of importance to the study was that the interview situation was
reasonably natural and that the participants talked as they do usually. This is not only a result of the general psychological, social,
and physical context, but also a result of whom they are talking to.
The conversational partner needs to give proper feedback in order
for the conversation to run smoothly, and this was done.
The influence of camera anxiety is a factor that could detract
from the naturalness of the situation. It could make participants
remember fewer details. Camera anxiety could lower the number of
gestures participants use. However, this effect would be expected to
influence across all conditions equally, and is therefore not a problem.
There are indications that the situation was natural. All participants used gestures. Indeed, it was found that all participants used
every type of gesture from the coding scheme, except a type of gesture called emblems, which is relatively uncommon anyway when
talking about past events (McNeill, 1992). None of the participants
commented on their own gestures.
132 Participating in a story
5.5.4 Possible effects of re-playing the game
It turned out that participant 3 had already played Anchorhead some
years before. What possible influences may the fact have that he was
re-playing the game? His motivation could be lower, because he was
not as curious about the game. This could in turn lead to lower encoding in memory. The consequences would be less memory for
detail, but would not likely affect conceptualisations. Low motivation could also make him fail to complete the required part. However, he completed it. Indeed, his motivation may be higher because
he got to re-experience something he knew he liked before. As an
indication of high motivation, he stated at the beginning of the
study that he wished to play it again. He spent considerably less
time than the other participants completing the required part (see
Table 5.5). In the interview, he made some statements that suggest
he knew details of the game from earlier sessions, as shown in examples 5.1 and 5.2.
(5.1)
Participant 3 remembers details from an earlier session (Interviewer,
Pierre):
Paul: där det låg en madrass
where there was a mattress
Pierre: mm
Paul: så den var mystisk
so it was mysterious
Pierre: mm
Paul: den tänkte man att man skulle ha till nånting
you thought that you would use it for something
fast det skulle man inte (.) det kommer sen
but that was not the case (.) that comes later
(5.2)
Participant 3 remembers details from an earlier session:
det fanns en en liten gränd också vid stadshuset som
there was a small alley too at the courthouse that
var bara en dead end (.) å den vet jag vad man har till
was just a dead end (.) and I know what that is for
fast det vet jag för att jag spelat det förut
but I know that because I have played it before
A method for generating data 133
Another effect could be that he remembered the spatial layout of
the game world better. However, from the interview data, he shows
signs of considerable loss of memory for places, which suggests his
memory from when he played it earlier could not have helped him
much when re-playing, as shown in example 5.3.
(5.3)
Participant 3 (Paul), Interviewer (Pierre)
Paul: e jag gick inte omkring i huset
eh I did not walk around in the house
Pierre: okej
okay
Paul: utan jag följde med michael upp för trappan direkt (.) så det
instead I went up the stairs with michael right away so it
var (.) foajé trappa upp och sen rakt fram var
was (.) foyer staircase up and then straight ahead was the
masters bedroom
master bedroom
[så att] (.) det gjorde jag direkt (.) så jag vet ingenting
[so that] (.) that I did right away (.) so I know nothing
Pierre: [mm]
Paul: om huset i övrigt (.) jag kommer inte ihåg faktiskt
about the rest of the house (.) I don’t remember actually
hur det såg ut ()
what it looked like ()
Pierre: mm
Paul från när jag spelade förut
from when I played before
In summary, the fact that participant 3 had played the computer
game before does not seem to have had substantial effects. It could
lead to better memory for details, although there was an indication
that the spatial knowledge may not be affected. His data were kept
and analysed in the same way as the other participants’ data.
5.5.5 Generalisability
A final point in this chapter is the question to what extent the data
collected in this study can be generalised to other people, stories,
134 Participating in a story
and situations. This depends on what kind of analysis is made and
not much can be said on a general level. There is nothing, however,
that would stop generalisation from the participants to the population in mind for this study: narrative-experienced and interactivefiction-experienced adults in a Western culture. When it comes to
the work of interactive fiction used in the study, it is representative,
and results should apply directly to other works of text-based interactive fiction. Considering other types of computerised participatory stories, such as those with audio and video, generalisation
could be affected (for instance, participants’ strategies for giving
spatial descriptions may depend on the viewpoint used in the representation of space in the participatory story). This depends on what
analysis is carried out, and the possibilities for generalisation will
consequently be discussed in the appropriate chapter. The case is
similar when considering even more different types of participatory
stories, such as live role-playing.
This chapter sketched the context of interactive fiction and discussed and justified the methodological decisions made for the collection of data. Chapters 6–8 are concerned with various analyses of
the collected data, starting with spatial cognition.
6
CHAPTER 6
Long-term memory representations of
spatiality24
as well as many other species in the animal
Wkingdom, move,around
in the world they gather information
HEN HUMANS
about the spatial layout of their surroundings and construct a mental representation of the spatial world (Tolman, 1948; Taylor &
Tversky 1996). This knowledge is sometimes called a cognitive map
(Kitchin & Freundschuh, 2000). The scale of such spatial relations is
that of large-scale, geographical space (Mark, Freksa, Hirtle, Lloyd,
& Tversky, 1999), or environmental space (Freundschuh, 2000), as
opposed to other spaces, such as the much smaller space around the
body, or very large spaces like astronomical spaces. Here, the term
spatial mental representation will be used for the knowledge a person
has, in long-term memory, about spatial relations (knowledge refers
to information believed to be true by the person, not necessarily
actually being true). It is not only from direct experience of surroundings that people form spatial mental representations, but also
from viewing maps and listening to and reading verbal descriptions.
How do people remember and think about space in a participatory story? A piece of interactive fiction such as Anchorhead can be
said to contain a fictive spatial world, a game world, that is, locations
which are related spatially, much in the same way as the fictive spatial world in a traditional novel, for instance, a town where the
story is set. Research on reading has shown that readers normally
do not construct mental representations in long-term memory of
24
Parts of this chapter have been published in Gander (2004).
135
136 Participating in a story
spatial relations described in text unless the task demands it (Zwaan
& van Oostendorp, 1993; Hakala, 1999). Reading a novel for entertainment would be a case where readers normally do not construct
spatial mental representations in long-term memory. The reader’s
task determines what kind of inferences are made on-line during
reading (van der Broek et al, 2001) and what kind of mental representations are created (Wilson, Rinck, McNamara, Bower, & Morrow, 1993).
The results obtained in studies of reading of ordinary texts raise
the question: Is spatiality spontaneously encoded in long-term
memory by players of interactive fiction? Spatiality is sometimes
noted to be a central property of participatory stories (Murray,
1997). Considering the nature of interactive fiction such as Anchorhead, there is reason to believe that the construction of spatial
representations differs from traditional texts. In these works it is
necessary for the player to move around in the game world in order
to make progress (to ‘consume’ the work). This navigational demand may drive players to construct spatial mental representations.
On the other hand, it could be that players move around in the
game world randomly, without forming any global spatial mental
representation. Or, a more situated scenario may be the case: Players could read descriptions of directions available in the text at each
place locally, and use this information to navigate, avoiding the
need for any global spatial mental representation. Research on reading of traditional texts also shows that readers are unlikely to construct spatial mental representations when the narrative is indeterminate, poorly organised, or has overly difficult descriptions (Denis
& Denhière, 1990; Mani & Johnson-Laird, 1982; Perrig & Kintsch,
1985). Interactive fiction has often been described as an indeterminate and fragmented narrative. It often seems that coherence is missing in the textual fragments that make up the physical object of a
work of interactive fiction. If this is the case, it would be another
reason to suggest that players perhaps do not form spatial mental
representations.
The question of spatial mental representations in interactive
fiction has not been thoroughly investigated. Tromp (1993) studied
Long-term memory representations of spatiality 137
the experience of spatiality and performance of users of M U D s
(multi-user dungeons). M U D s are very similar to a typical work of
interactive fiction, the main difference being that many players are
active at the same time in the game world. Using an electronic questionnaire, she concluded that M U D players construct elaborate and
highly accurate spatial mental representations. Dieberger (1994)
interviewed users of a M U D about their opinions on how spatiality
was designed in the M U D . However, both studies relied on players
introspecting on their abilities and memories, and it would be desirable to further study the question of spatial mental representations
using more valid methods.
The present study of spatial cognition is outlined as follows.
Parts of the interview data collected from eight participants (see
Chapter 5) concerned spatiality. Participants who played the computer game were asked if they could tell which places were present
in the game world, and what the relative locations of these places
were. To give an opportunity for comparison, the same questions
were put to four other participants about the spatiality of a nonparticipatory version of the computer game. The non-participatory
version was an adapted version of an actual game session (see Chapter 5). The spatial game world was the same as the one in the game.
The printout of the non-participatory version made it in some sense
similar to a traditional short story. In one sense, it functions as a
direct case of comparison against the computer game. However, the
non-participatory version was somewhat artificial. For instance, the
text was written from the unusual perspective of second-person singular—you—a property that sets it apart from most authentic short
stories. Since only the computer game and the non-participatory
version contained similar spatial descriptions, the analysis in this
chapter is limited to these two conditions. Participants 1–4, who
played the computer game, will be referenced in the following as
players, and participants 5–8, who read the non-participatory version will be called readers.
During the collection of data for the present study, no instructions to memorise the spatial layout of the game were given, and it
is unlikely that participants suspecting a memory test memorised
138 Participating in a story
the spatial layout. The first four participants simply played the computer game in the way they would normally do it (see Chapter 5 for
details). In contrast, the non-participatory version was somewhat
artificial, and the other four participants could have been prepared
for a memory test. However, as was seen from the data, readers remembered less of the spatial layout than the players, which makes it
unlikely that readers anticipated a memory test.
The data on spatiality were analysed with regard to two separate
but connected themes. First, an analysis was made of how people
talked about space from interactive fiction. To study speech is one
way of approaching the question of spatial mental representations.
A series of questions guided the analysis of the interview data of
players of the computer game and readers of the non-participatory
version of the computer game: What terms—referents and verbs—are
used when talking about space? How are spatial references made?
What perspectives are used when giving descriptions of spatiality
from the game?
In the second theme on spatiality, the focus is shifted from the
descriptions of spatiality to the spatial mental representations themselves. The first question to be considered is what the order in
which participants describe places can tell us about the spatial mental representations. Second, to what extent do players’ and readers’
spatial mental representations reflect the true state of affairs? This
question is examined by looking at the completeness, accuracy, consistency, and integration of the spatial mental representations. The
property of hierarchical organisation of the spatial mental representations is also investigated. Finally, to the extent that the participants make mistakes, the question whether these errors are systematic is explored. For instance, is there evidence for distortions that
are common in memory for graphical maps, such as alignment and
rotation?
Next, an analysis of how the participants talk about space in the
game world is presented.
Long-term memory representations of spatiality 139
6.1
TALKING ABOUT SPACE FROM INTERACTIVE
FICTION
How do we express in language our recollection of some objects
located in space? In Levelt’s (1996) model, describing a spatial scene
requires selecting a referent, a relatum (what it is to be related to),
their spatial relation and also applying some perspective system that
will map spatial relations onto lexical concepts. Information that
resides in the mind’s many representational systems, for instance,
the spatial representational system, needs to be translated into a
semantic code that can be formulated in the language. Levelt
identifies three perspective systems: relative, intrinsic, and extrinsic.
The relative system is deictic and relative to the describer’s current
position and orientation, for example, ‘the pig is to the left of the
tree’. The intrinsic perspective system relies on the relatum’s intrinsic axes, for example, ‘the cow is in front of the horse’, where the
horse has an intrinsic front. Finally, the extrinsic perspective system
is an absolute system, for example, ‘the cat is north of the house’.
The choice of perspective system is linguistically free (Levelt, 1996);
that is, there is no unique or fundamental choice. It is not biologically determined and cultures have ended up using just one or a mix
of the perspective systems. For instance, on the island of Vava’u in
Polynesia, the inhabitants mainly use an absolute frame of reference
(Bennardo, 2002). English uses all three perspective systems, although the relative perspective system is favoured (Li & Gleitman,
2002). The choice of perspective system can also depend on individual factors as well as on the task at hand.
When giving verbal descriptions of a spatial arrangement of
landmarks, such as a landscape, a town, or a building, people can
use any of three descriptive strategies, called route, gaze, and survey
(Taylor & Tversky, 1996). These descriptive strategies are linked to
Levelt’s perspective systems. A route description takes the addressee
on a mental tour of the surroundings, commonly using a relative
reference system, such as ‘if you turn left and walk two blocks, you
have a big red house on your right’. A gaze description is given
from a fixed position outside the scene and usually describes a small
140 Participating in a story
sized area, for example, ‘the bed is to the left of the window’. Finally, a survey description is made from above, using an extrinsic
reference system, for example, ‘the church is north of the forest’.
Although it was first claimed that people select one descriptive
strategy and keep it fixed throughout the description, later research
has shown that people often switch strategies within one and the
same description, usually without signalling the switch (Tversky,
Lee, & Mainwaring, 1999).
The purpose of this first part of the chapter is to investigate how
players and readers talk about the spatiality of the game world. The
main motivation was to study the descriptive strategies of the players, and compare these with descriptive strategies found in earlier
research of people learning spatial layouts in other situations. The
readers serve as control group when the participatory property is
left out.
6.1.1 Analysis
In analysing how players talk about space in the computer game,
the procedure here follows the outline of Taylor and Tversky
(1996). First, the overall perspective was judged for each description
made by the participants (route, survey, or gaze, or a mixture of
these). Then, a more detailed analysis of language use was made on
three types of elements which make up the perspective: what relational terms were used (e.g., left, north), what referents were used
(e.g., the addressee, landmarks), and the occurrence of active and
stative verbs (e.g., enter and to be, respectively).
6.1.2 Results
Below, the results of analysis are presented for descriptive perspectives, relational terms, referents, and verbs, and under each point, for
the players followed by the readers. For players, data from three
participants were analysed (participants 1–3) (data from the fourth
participant on an explicit instruction to describe places was missing). For the readers, data from four participants were analysed
(participants 5–8). Descriptions were available for the overall layout
Long-term memory representations of spatiality 141
of the game (referred to as City) as well as for a smaller part of the
game (referred to as House).
6.1.2.1 Descriptive perspectives
Following Taylor and Tversky (1996), first, each landmark description in the descriptions was coded according to perspective type
(route, survey, or gaze). Landmarks were mainly locations (places
possible to move between in the computer game), but sometimes
what was technically a single location had multiple landmarks (e.g.,
the square, which was a landmark in itself, also had an obelisk in its
centre—a landmark but not a location). Then, depending on the
landmark coding, the entire descriptions were classified into route,
survey, or gaze if at least all but two landmarks were described using that perspective. Descriptions were classified as mixed if two or
more descriptions of landmarks were classified as belonging to a
perspective different from the main perspective.
Players. The results (Table 6.1) show that participants use survey
descriptions exclusively (i.e., an extrinsic reference system using
north, south, etc.) when describing the spatial layout of the game
world. All descriptions were started at the same place, Outside the
real estate office, where the game begins.
Readers. The results show that readers mainly used survey descriptions when describing the spatial layout of the game world
(Table 6.2). In some cases, the participants did not relate landmarks
spatially at all. Two descriptions were started at the location where
the game begins (Outside the real estate office), while three descriptions had other starting places (The pub, The house, and The university).25
6.1.2.2 Relational terms
Players. In the descriptions made by players, only extrinsic relational terms (e.g., north, south) were found, shown in the left part
of Table 6.3.
25
One participant gave two separate descriptions, thus the total sum of five.
142 Participating in a story
Table 6.1. Occurrence of perspective in the descriptions of the computer game for
participants 1–3.
Spatial area
City
House
Total
Route
0
0
0
Gaze
0
0
0
Descriptive perspective
Survey
3
3
6
Mixed
0
0
0
Table 6.2. Occurrence of perspective in the descriptions of the game world for participants 5–8.
Spatial area
City
House
Total
Route
0
0
0
Gaze
0
0
0
Descriptive perspective
Survey
Mixed
3
0
3
0
6
0
None
1
1
2
Note: None means no description strategy could be determined because no spatial
relations were present in the data.
Table 6.3. Frequencies of all relational terms from the participants.
Players
Relational term
Occurrences
south
14
north
9
down
8
up
7
west
6
southwest
5
east
4
close to
3
southeast
2
northwest
1
Readers
Relational term
Occurrences
east
4
north
1
up
1
west
1
Note: In the data, up (upp) was found to be used similarly to above, and is treated as
an extrinsic relational term, although it is strictly relative to, e.g., the body.
Long-term memory representations of spatiality 143
Two occurrences of fram (forward) were found. The first one
was komma fram till (arrive at), which suggests an orientation towards the thing that one arrives to. However, since this was a single
isolated case, it was not considered further. The other occurrence
was rakt fram var masters bedroom (straight ahead was the master
bedroom) (see the discussion section below).
Readers. In the spatial descriptions made by readers of the nonparticipatory version, relational terms were sparse. It was more
common to name locations without giving a spatial relation than to
relate locations spatially. The few relational terms that occurred
were extrinsic, shown in the right part of Table 6.3.
6.1.2.3 Referents
Players. In talk about spatiality from the computer game, the occurrences of referents were about equally split between using a landmark as referent and using canonical directions (north, east, etc.).
For the landmarks, a mixture of English and Swedish terms was
used, even within one and the same participant and within one and
the same utterance.
Readers. In the non-participatory version, landmarks were used as
referents by all four participants. In the description of one participant, six occurrences of canonical directions were present.
6.1.2.4 Verbs
Players. For players, the stative verbs were about twice as numerous
as the active verbs, as shown in the left part of Table 6.4. The class
of active verbs, used to describe, for example, the path of a road,
called fictive motion by Talmy (1996) had only a single occurrence
in the descriptions (gå/go), as shown in example 6.1 (see Table 5.3
in Chapter 5 for a transcription key).
(6.1)
Participant 2:
därifrån nånstans så kom man till en väg som
from somewhere around there one came to a road that
gick (.) i nord sydlig riktning
went (.) in a north south direction
144 Participating in a story
The relational term up always occurred together with an active
verb, the single exception being där uppe låg huset (up there was the
house).
Readers. Stative verbs dominated the descriptions from the readers, as shown in the right part of Table 6.4.
6.1.2.5 Summary of results
A summary of the results of spatial descriptions and comparison
between computer game and non-participatory version is shown in
Table 6.5. In the table, dominant means most frequently occurring,
which serve as the basis for classifying descriptive perspectives (Taylor & Tversky, 1996).
Table 6.4. Number of occurrences of stative and active verbs from participants’
descriptions.
Players
Stative
verbs
Readers
Active
verbs
Stative
verbs
Active
verbs
finnas (be)
22
gå (go)
21
vara (be)
11
vara (be)
16
klättra ner
1
ligga (lie)
10
ligga (lie)
5
(climb down)
leda (lead)
1
gå (go)
3
Note: English translation within parentheses.
Table 6.5. Summary of the results of spatial descriptions and comparison between
computer game and non-participatory version.
Source condition
Topic of spatial description
Computer game
Dominant descriptive perspective
Survey
Non-participatory
Survey
Dominant relational terms
Extrinsic
None
Dominant referents
Landmarks,
canonical directions
Landmarks
Dominant verbs
Stative
Stative
Long-term memory representations of spatiality 145
6.1.3 Discussion
6.1.3.1 Descriptive strategies, relational terms, referents, and verbs
Descriptions of the spatial layout of a work of interactive fiction
were analysed with respect to reference system, referents, and verbs,
in order to determine whether a gaze, route, or survey description
was used. The results show that players as well as readers used a
survey description throughout the descriptions: The relational
terms were all extrinsic, referents were either landmarks or the canonical directions and the dominant verb class was stative verbs—all
characteristics of a survey description (Taylor & Tversky, 1996).
The addressee (i.e., the interviewer) was never used as a referent
(i.e., no participant used the interviewer as a spatial referent by saying, e.g., ‘to your right’). In one description, there was one possible
instance of one landmark being described according to a route
description, shown in example 6.2.
(6.2)
Participant 3:
utan jag följde med michael upp för trappan direkt (.) så det var
but I followed michael up the stairs directly (.) so it was
[(.) foajé]
[(.) foyer]
[((both hands shape an enclosed space: iconic gesture, observer viewpoint))]
[trappa upp] och
[up a staircase] and
[((right hand moves up: iconic gesture, observer viewpoint))]
[sen rakt fram var masters bedroom]
[then straight ahead was the master bedroom]
[((right hand open palm fingers together back a bit then
forward: iconic gesture, subjective viewpoint))]
This occurred in the description of the house. This fact could have
influenced the switch to route perspective, since a small environment influences the description towards route descriptions (Taylor
& Tversky, 1996). However, an alternative explanation could be
that ‘forward’ might mean ‘north’, considering that north is up
146 Participating in a story
(forward) on a two-dimensional map (the influence that the game is
to be thought of as such a map is very strong—using the reference
system of a map, and within the genre of these kinds of games it is
frequent to draw these maps). In support of this idea, ‘down’ was
sometimes used to mean ‘south’, such as in example 6.3 from participant 1.
(6.3)
Participant 1:
uppdelad i två bitar eh jag man började i den norra bidivided into two parts eh I you started in the north paden norra delen (.) o:ch sen så skulle man gå ner till söder
the north part (.) a:nd then you were supposed to go down south
The players used the canonical directions (north, south, etc.) as
landmarks to a greater extent than the readers. The reason for this
may be that players to a large extent used the canonical directions
for movement when playing the game—this is how most of the
movement in the game works. But it could also be that the players
were more prone to think of spatial issues because of the navigational demand, and this made them more aware of how the game
world was configured in terms of the canonical directions. As will
be seen in the second part of this chapter, the players had superior
knowledge of the spatial layout, compared to the readers.
Interestingly, active verbs were commonly used in the survey
perspective description using an extrinsic reference system, and also
more so for the players than for the readers. The use of active verbs
is not common in spatial descriptions made from a survey perspective (Taylor & Tversky, 1996). The explanation for this could be
that the players performed movements in the game, and this made
the active verbs more salient than for the readers, who did not focus
in the same way on the movements described in the text they read
(the non-participatory version).
6.1.3.2 Why not a route perspective?
Why did players not adopt a route perspective when describing the
environment? This would be expected given that players learned the
Long-term memory representations of spatiality 147
environment by exploration. Taylor and Tversky (1996) obtained
experimental results that when learning an environment by exploration (in contrast to learning it from descriptions or from maps),
people are more likely to adopt a route perspective in their descriptions of that environment. However, there are at least three factors
present in the current situation against using a route description.
First, the choice of perspective has been shown to be influenced by
certain features of the environment described. If the environment
features a single path or has landmarks of roughly the same size, it is
more likely to be described using a route perspective (Taylor &
Tversky, 1996). The game world, in contrast, has multiple paths and
feature landmarks of various sizes (e.g., a bathroom, a deserted lane,
a square). Second, the conventional use of extrinsic referential terms
(north, south, etc.) within the genre of interactive fiction most
likely influenced participants’ choice of perspective. The fact that
the game used an extrinsic reference system may have forced participants into adopting a survey perspective. Third, related to this, is
the fact that in the game, ‘you’, the player character, cannot change
orientation. ‘You’ move from place to place, always facing the same
direction. There is no information of a particular direction in the
location descriptions from the game. Even if orientation changes are
sometimes implied, as when a description is given of possible paths
of movement, the orientation is always returned to an unspecified
state. The following example of a description of a location from the
game illustrates this point:
Outside the Real Estate Office
A grim little cul-de-sac, tucked away in a corner of the claustrophobic
tangle of narrow, twisting avenues that largely constitute the older
portion of Anchorhead. Like most of the streets in this city, it is ancient, shadowy, and leads essentially nowhere. The lane ends here at
the real estate agent’s office, which lies to the east, and winds its way
back toward the center of town to the west. A narrow, garbagechoked alley opens to the southeast.
Further, there are at least two major methodological differences
between the present study and that of Taylor and Tversky (1996).
148 Participating in a story
One difference concerns the task instructions. In their study, participants were told to give descriptions ‘so that someone who was
unfamiliar with the environment and had never seen the map could
read the description and know where all the landmarks were’ (Taylor & Tversky, 1996, p. 378). This may have influenced Taylor and
Tversky’s participants to use a route perspective more frequently in
order to guide another person around the landmarks. A second difference is that the participants supplied their descriptions in writing,
while in the present study, participants used speech. It is unclear how
spatial descriptions are affected by this difference, but it is conceivable that it may have an effect on perspective choice, especially since
this choice takes place in the later stages of language production
(Levelt, 1996).
6.1.3.3 Starting points of the descriptions
Even though the players gave survey descriptions of the computer
game, they all started their descriptions with the same landmark—
the place where the game begins. Among the readers, it was as likely
to start descriptions at the starting point of the game as at some
other significant location. The reason for this difference may be that
players see the starting point as more important, since they worked
on the puzzle in the game of how to enter the office. Both kinds of
starting points are consistent with Taylor and Tversky’s (1996) observation (although they were speaking about route descriptions)
that starting points typically were entrances or large landmarks
(‘large’ is translated to ‘important’ in the present study).
6.1.3.4 Single perspective
Participants used a single perspective throughout their descriptions.
In contrast, Taylor and Tversky (1996) found that perspective
switches were frequent. Apparently, in the present study, participants felt they could adequately describe the environment with a
single perspective. This suggests that the environment was perceived
as homogenous by the participants or that perspective switches
Long-term memory representations of spatiality 149
would require too high cognitive effort to be worth while (Tversky,
Lee, & Mainwaring, 1999).
A final question is what descriptive strategies tell us about the underlying long-term memory representation of spatiality. According
to Levelt (1996), it tells us very little, since perspective is not part of
the mental representation of space but is chosen at a late stage in the
verbalisation process. But the study of descriptive strategy does tell
us two things. First, as has been discussed, the descriptive strategy
used by the participants in the present study tells us something
about how the spatial descriptions are made when a spatial layout
was learned in a work of interactive fiction. This case can be compared against other cases, such as learning from maps, from written
instructions, from exploration, etc. So, even if descriptive strategies
do not tell us directly about the memory representations, they do
reflect parts of the ongoing cognitive processes at the time the description was given. Second, the fact that players—in contrast to
readers—gave elaborate spatial descriptions suggests that they were
indeed drawing their descriptions from long-term memory spatial
mental representations. The following section investigates this issue
in detail.
6.2
SPATIAL MENTAL REPRESENTATIONS
In this second part of the chapter, the focus is the nature of the
mental representations of spatiality constructed by participants. But
before proceeding, there is a need to discuss the fundamental question of whether people form spatial mental representations in longterm memory at all.
6.2.1 Do people form spatial mental representations?
Two lines of criticism of the claim that people construct spatial
mental representations are that (i) spatial representations are not
150 Participating in a story
truly spatial, and (ii) spatial representations are spatial, but they may
not be constructed in a given situation.
The first line of criticism concerns the very nature of mental representation and questions whether spatial representations are truly
spatial. Even though everyone may agree that there are mental representations of spatial facts about the world, one can disagree on the
question whether the representations themselves are actually spatial.
After all, even mental representations in the form of propositions
can represent spatial facts about the world, without themselves being spatial in nature. The issue of mental imagery has had a long
history of debate within the field of cognitive psychology. Note,
however, that image is not synonymous with spatial mental representation, because the latter has non-image-like features, such as being non-metric and combining information from several sensemodalities (Tversky, 2000). For the purpose of this study, however,
this fundamental question bears little relevance. The question does
not address the issue of differences in mental representations of spatial facts about story texts and interactive fiction. If one claims that
spatial mental representations cannot be spatial, then this holds irrespective of whether the mental representations are about stories in
traditional media (such as novels and films) or interactive media
(such as interactive fiction). If one accepts that mental representations can be spatial, then this of course is a possibility for both
cases.
Turning to the second line of criticism, even agreeing that spatial
mental representations are possible, there is still the question of under what circumstances people actually construct spatial mental
representations. Taylor and Tversky (1992) found that when given
descriptions of fictive places to read, without explicit instructions to
remember the spatial layout, people did construct spatial mental
representations spontaneously. However, this conclusion can be
criticised on methodological grounds. It was an artificial laboratory
situation, and not a real story. Even if participants were not told
they were later going to draw a map, because of the situation they
could have anticipated a memory test of the contents of the text. As
was discussed above, it has been found that readers of normal texts
Long-term memory representations of spatiality 151
in naturalistic settings do not normally construct long-term memory representations of spatial relations. Nor do readers construct
spatial mental representations when the narrative is indeterminate,
poorly organised or contains overly difficult descriptions. The present study is arguably naturalistic when it comes to playing the
computer game (see Chapter 5). So, do players construct spatial
mental representations of the game world in normal game playing,
and what evidence would let us determine this?
6.2.2 Determining spatial mental representations
Starting off with audiovisual data on players talking about the computer game after having played it, there are several types of evidence
that would support the claim that participants form spatial mental
representations of the game world in long-term memory:
(i)
Elaborate spatial descriptions. The basic observation that players
actually gave lengthy descriptions, spatially relating various
places from the game is intriguing. This observation does not
tell us whether any spatial mental representations reflect the actual spatial layout of the game, but even taken alone, it makes
it probable that players drew their spatial descriptions from a
spatial mental representation—truthful or not.
(ii) Description order. If the order in which places from the computer game are mentioned by the participants matches the order in which the places were visited, participants may simply
recall places off a temporally ordered list that was memorised
at the time of encoding. This would suggest that no spatial
mental representation was formed. On the other hand, if participants mention places that are spatially proximate when considering the spatial layout as a graphical map—in other words,
an order that is consistent with doing a sweep across a map—
there is reason to believe that a spatial mental representation is
the source of these descriptions.
(iii) Completeness and accuracy. Another piece of evidence is correspondence of the talk about spatiality to the true spatial layout
152 Participating in a story
of the computer game by the description being complete and
correct.
(iv) Consistency and integration. A fourth piece of evidence is if
descriptions present a consistent and integrated picture of the
spatial layout. In other words, it is suggestive of spatial mental
representations if participants do not contradict themselves
when giving spatial descriptions and if the descriptions present
a connected whole.
(v) Other features of spatial mental representations. The descriptions
should express properties that spatial mental representations
have been shown to have in other studies, such as hierarchical
organisation. The presence of systematic distortions that is
common in memory for graphical maps would also be revealing.
The first point has already been discussed in the first part of this
chapter. The last four points will now be investigated by analysing
the verbal descriptions of spatiality made by the participants.
6.2.3 Method
6.2.3.1 Analysis of description order
For each participant, a list of places from the computer game world
(henceforth called locations) was constructed, in the order they were
mentioned in the descriptions. Synonyms and translated names of
locations were considered in order to resolve referential ambiguities.
References that were still uncertain were marked as such. Some
place names could not be mapped directly to locations in the game.
These were listed in italics by the name used by the participant in
the description. For each of the players, a second list of locations
was constructed. The lists were arranged in the order in which the
participants visited the locations, resulting in a unique list for each
player. For the readers, a single list was made from the order the
Long-term memory representations of spatiality 153
locations were mentioned in the text of the non-participatory version of the computer game.
From the same transcriptions, for each player, a graphical map
was constructed by interpreting the participant’s description using a
set of drawing conventions, shown in Table 6.6. The graphical maps
thus represent the spatial information in each player’s description.
The purpose of constructing the graphical maps was twofold. First,
the maps were used when comparing the mentioned order of locations to see whether description order matched spatial proximity.
Second, the maps were used when investigating the truthfulness of
the descriptions by comparing them to actual maps from the computer game. Note that the constructed graphical maps did not represent distance, because this information was not available from the
descriptions (variation in distance on the constructed maps was
made according to drawing convenience). One possibility to reconstruct distance would have been to use participants’ choice of verbs
as an index of distance, for example, ‘walk’ would mean a short distance, and expressions such as ‘is located north of’ would mean a
long distance. However, this procedure was abandoned, mainly for
its speculative nature, but also because distance is not represented in
the original map anyway (in the game, ‘going west’ or ‘travelling
west’ moves the player the same distance, always one location at a
time).
As a final step, the order of mentioning the locations was compared to the order in which the locations were visited as well as to
spatial proximity on the constructed graphical maps.
154 Participating in a story
Table 6.6. Drawing conventions used when constructing graphical maps from participants’ descriptions.
Description element
Drawing convention
Relation
Arrow labelled in capital letters with relational term
used
(no relation mentioned)
Locations mentioned but not related spatially by the
participants were drawn in separate, spatially unconnected solid-line boxes
Hierarchical regions
Dotted-line ellipses
North
Above
East
To the right
South
Below
West
To the left
Northeast
Above to the right
Southeast
Below to the right
Southwest
Below to the left
Northwest
Above to the left
Up
Above, labelled ‘U P’
Down
Below, labelled ‘D O W N’
In
To the right, labelled ‘I N’
Out
To the left, labelled ‘O U T’
Note: Additional relations, not in the list, were labelled on the map with the term
used by the participant and positioned in ways most practical to fit on the map.
6.2.3.2 Analysis of properties of spatial mental representations
In order to compare how well the given spatial description actually
corresponded to the true spatial layout of the game world, four
measures were employed. Completeness of the spatial mental representation was calculated by the number of described locations divided by the number of actually visited locations (to count, the
mentioned locations had to be locations actually occurring in the
game). Accuracy was calculated by number of correct descriptions of
spatial relations between two locations divided by the total number
of descriptions of spatial relations between two locations. Consistency is judged by the absence of contradictions. A contradiction
Long-term memory representations of spatiality 155
was said to have occurred if the participant gave incompatible spatial relations between two locations, for example, first saying the
pub was north of the church, then later saying that it was south of
the church. A low number of contradictions suggest a stable mental
representation of the spatial layout. Integration concerns the question of whether the participant’s description of the spatial layout
forms a connected whole, or if the description consists of a number
of separate, spatially unconnected regions. A region was defined as
at least two locations related spatially.
Hierarchical organisation was investigated by looking for superordinate categories in a part-relationship to location categories.
For instance, a forest would be a superordinate category to a clearing, and a hospital would be a superordinate category to a waiting
room (subordinate categories were not considered, since locations
are instances of categories, and thus already at the ‘lowest’ level).
The participant’s descriptions were searched for errors, and these
were compared to known distortions for graphical maps, such as
alignment bias (Tversky, 1981).
6.2.4 Results
Here, the results are presented concerning the analysis of description order, completeness, accuracy, consistency, integration, hierarchical organisation, and systematicity of errors.
Participant 4 was not explicitly asked to describe places and their
locations, as were the other participants, but nonetheless gave
lengthy spontaneous descriptions when asked a different question.
These data were analysed in the same way as for the three other
participants, but one should keep in mind that the circumstances
under which the data for Participant 4 were collected differed from
the others.
6.2.4.1 Description order
Players. The order in which locations were mentioned in the descriptions made by the players sometimes, but not always, matched
the order in which the locations were visited. In the same way,
156 Participating in a story
mentioned order sometimes, but not always, was consistent with
scanning a graphical map of the locations.
Participant 1 gave three consecutive descriptions, with each description starting over from the beginning, containing varying
amount of details in each version. Participants 2 and 3 gave two
separate descriptions each. It is the order within each of these descriptions that was compared to the order in which the locations
were visited. There were insufficient data for participant 4 from
which to construct a list. An example of a list of locations in visited
order and mentioned order can be seen in Table 6.7 (for participant
2).
Figure 6.1 shows an example of a graphical map constructed from
descriptions (for participant 2).
Mentioned order was compared to visited order and spatial proximity on the maps for each participant. Participant 1 made four descriptions. The first three describe the overall layout of the game,
and can be considered three versions of the same description (the
participant can be seen starting over from the beginning in each
description). The fourth description is of the House. In the first
short description of places, the order of mentioning is the same as
the order visited. In the second, longer description, the mentioned
order of descriptions partly matches the order visited. One exception is the occurrence of University court and Library, which appear
much earlier in the description than in the order visited. This suggests that the description was generated by some other means than
going through the locations in the same sequence they were visited
while playing the game. Comparing the list of mentioned locations
with the constructed map for participant 1, it was seen that the description order was consistent with doing a mental sweep on a spatial representation from east to west. The other exception is the
occurrence of The house before the occurrence of The sea, with the
latter being added at the end. One explanation for this could be that
the participant inspected the spatial mental representation when
talking about The house, and found that, around that area, The sea
was located. The last exception is the mentioning of Town square
(although this reference is uncertain; the participant uses the term ‘a
Long-term memory representations of spatiality 157
square’), which is not consistent with the order visited. It may be
represented close to Twisting lane in the participant’s spatial mental
representation, and in that case it is consistent with reading off a
spatial representation, although it is not located in the place described by participant 1 on the objective game map.
The order of locations in the description made by participant 2
roughly followed the order the locations were visited. Two positions in the orderings are sometimes switched, sometimes three.
Once exception is that University court is mentioned last, after The
house. On the map, these two places are located at two separate
ends. This makes it unlikely participant 2 came to focus on University court by reading off a spatial mental representation.
For participant 3, the order mentioned does not match the order
visited for the most part.
Table 6.8 summarises the results of matching the mentioned order to visited order and spatial proximity. Players are listed in the
upper half of the table. There was an artefact of the matching
method: As can be seen in the table, in the cases where a description
consisted of few locations (say, below 5), matches were very frequent.
Readers. The order in which readers mentioned the locations
generally did not match the order in which they occurred in the
text, as shown in the bottom half of Table 6.8. An exception was
participant 8 who, while mentioning the highest number of locations, also did so in the order they were visited. No graphical maps
were constructed for participants 5–8, since the spatial relations
were either very vague or missing completely. Thus, a graphical
map would in those cases be just a list of locations. Because of this,
it was not possible to compare mentioned order to maps.
158 Participating in a story
Table 6.7. For participant 2, the order in which the locations were mentioned (the
participant gave two descriptions), and the order in which the locations were visited.
Locations
Order mentioned
First description
Second description
The real estate office
Narrow beach
Local pub
Town square
Whateley bridge
Asylum courtyard
Court house
Under the bridge
Vacant lot
Wharf
Deserted lane
Churchyard
House
University court
Upstairs landing
Master bedroom
Bathroom
Foyer
Darkness ?
Order visited
Outside the real estate office
Alley
File room
Office
Narrow beach
Narrow street
Twisting lane
Local pub
Whateley bridge
Town square
Dark corner
Asylum courtyard
Riverwalk
Vacant lot
Wharf
Under the bridge
Chilly avenue
Deserted lane
Churchyard
Behind the church
Down the road
Side alley
Junction
University court
Library
Scenic view
Outside the house
Foyer
Upstairs landing
Darkness 1
Darkness 2
Master bedroom
Bathroom
Note: Question marks indicate unclear references. Italics mark terms used by the
participant but not matching any actual game location.
Long-term memory representations of spatiality 159
The University
Alley
Real estate office
‘didn’t get anywhere’
Beach
Pub
DOWN
Water
Under the bridge
Pipe
City centre
Asylum
Bridge
Obelisk
OUT
The
Square
Abandoned lot
Wharf
Road
Court house
Church
Churchyard
Fog
House:
The bathroom
The bedroom
Stairs
UP
The foyer
Figure 6.1. Graphical map constructed from participant 2’s description. See Table
6.6 for a list of drawing conventions used.
160 Participating in a story
Table 6.8. The results of matching the order in which locations were mentioned to
the order in which locations were visited/read and to spatial proximity on the constructed graphical maps.
Matches
Spatial description
Players:
Participant 1: Description 1
Participant 1: Description 2
Participant 1: Description 3
Participant 1: Description 4
Participant 2: Description 1
Participant 2: Description 2
Participant 3: Description 1
Participant 3: Description 2
Readers:
Participant 5: Description 1
Participant 5: Description 2
Participant 5: Description 3
Participant 6: Description 1
Participant 6: Description 2
Participant 7: Description 1
Participant 7: Description 2
Participant 7: Description 3
Participant 8: Description 1
No. of
locations
Visited/read
order
Spatial proximity
6
14
2
4
14
5
13
3
yes
no
yes
yes
no
no
no
yes
yes
yes
yes
yes
no
yes
no
yes
5
6
5
3
2
7
2
4
9
no
no
no
no
no
no
yes
no
yes
–
–
–
–
–
–
–
–
–
6.2.4.2 Completeness, accuracy, consistency, integration
Players. The left part of Table 6.9 reveals that the players’ descriptions of spatiality were relatively complete, to a high degree accurate, consistent and in most cases integrated. No locations that were
not in the game were described (i.e., there were no occurrences of
made-up or falsely remembered locations). It should be noted that
the results for participant 4 differ as a result of differences in the
eliciting conditions. The fact that participant 4 was not explicitly
asked to describe the spatial layout probably had the effect that the
description was less complete, accurate, and integrated than would
have been the case if this question had been asked explicitly.
Long-term memory representations of spatiality 161
Readers. The results of the readers present a contrasting picture,
as can be seen in the right part of Table 6.9. Readers gave descriptions that were mostly incomplete. Very few spatial relations were
mentioned, but the ones that did occur were about 50 percent accurate. Because of the low number of spatial relations, the consistency
and integration measures were not calculated for readers (see further
the discussion below).
Table 6.9. Completeness, accuracy, consistency, and integration of participants’
spatial mental representations of the layout of the game world.
Players
Measure
Readers
2
3
4
Mean
5
6
7
8
Mean
No. of described 14
locations
21
17
21
18
10
3
10
9
8
No. of encountered locations
26
33
26
42
32
25
25
25
25
25
Completeness
of locations
54% 64% 65% 50% 57%
40%
12%
40%
36%
32%
No. of accurate
relations
7
13
15
16
13
1
0
3
0
1
No. of de8
scribed relations
16
16
24
16
3
0
5
0
2
33%
–
60%
–
50%
1
Accuracy of
relations
88% 81% 94% 67% 80%
No. of contradictions
0
0
0
0
0
No. of regions
1
1
1
6
2.25
Note: The overall accuracy figure is lowered by the fact that the total number of
spatial relations consists of the categories correct, incorrect, and uncertain, the last
category including, e.g., spatial relations that may be correct but where no destination location was mentioned.
162 Participating in a story
6.2.4.3 Hierarchical organisation
Players. The players’ descriptions expressed hierarchical organisation
of elements (see Figure 6.1 for an example). Participant 1 talked
about the northern and the southern parts, and places were located
in each part. All players talked about the house and rooms within
the house. Participant 4 talked about the upper floor of the house
and what rooms it contained, expressing a two-level hierarchical
organisation. Participant 1 also used a two-level hierarchy when
talking about the rooms in the house, and said that the house was
located in the southern part.
Readers. Descriptions by readers also showed a hierarchical organisation. Participant 5 expressed a hierarchical organisation by
talking about the town and places within it. Participant 6 talked
about the first floor of the house, and said that the bedroom was
located there. Participant 7 also gave a similar description of the first
floor, and also described the town square and the places it contained. Participant 8 mentioned that the foyer and the master bedroom were inside the house.
6.2.4.4 Distortions and errors in spatial mental representations
Players. Although the players’ spatial descriptions were highly accurate, two kinds of systematic errors were found in the player’s descriptions: a possible case of alignment bias and confusions of the
canonical directions east and west.
The data do not permit analysis of fine-grained spatial differences
between the original map (the one constructed from the participant’s game session log) and the map as described (those kind of
analyses require distance information in an original map and a map
drawn by the person). However, the memory bias called alignment,
which causes objects in the memory of a map to be lined up horizontally or vertically (Tversky, 1981), could explain one inaccuracy
found in participant 2’s spatial representation. The location Vacant
lot was aligned with the road south which leads to The House. In the
original map, the Vacant lot lay more to the east, but this location
may have been seen as more important by the participant than the
Long-term memory representations of spatiality 163
Riverwalk which was actually situated in the place where Vacant lot
was described to be.
Another type of error was to confuse certain directions more often than others. The directions east/west were mistaken more often
than north/south. Participant 1 described an exit to the east when it
actually was located to the west, as shown in example 6.4.
(6.4)
Participant 1:
trappan ledde ju till sovrummet (.) å från sovrummet
the staircase led to the bedroom (.) and from the bedroom
fanns det två stycken utgångar en i söder å en (.) i öster
there were two exits one to the south and one (.) to the east
In example 6.5, participant 2 makes two errors. First, a beach is described as being located southwest instead of east or southeast. Second, the pub is described as being located east instead of west.
(6.5)
Participant 2:
(du) har ju fastighetskontoret (.) sen öh: (.) sen eh:
(you) have the real estate office (.) then eh: (.) then eh:
sydvä:st om de så fanns de nån strand (.) som man
southwe:st from that there was some beach (.) that you
kunde klättra ner på (.) där det fanns
could climb down onto (.) where there was
[en ett rör som stack ut]
[a pipe that stuck out]
[((shows an extended object with thumb and index finger three
times right-left-right: iconic gesture, unknown viewpoint))]
(.) °hh (.) å vatten då mm (.) lite mer ö:sterut (.) så fanns det en pub
(.)°hh (.) and water mm (.) a bit more to the ea:st (.) there was a pub
Speech repairs also suggest that east and west are confusing, as
shown in example 6.6.
(6.6)
Participant 2:
innan man gick ner dit så kunde man ta sig ut på en
before you went down there you could get to an
abandoned lot där det fanns en madrass och så (.)
164 Participating in a story
abandoned lot where there was a mattress and (.)
kunde man gå (.) söderut (.) sydväst! eller sydöst
you could go (.) south (.) southwest! or southeast
därifrån så kom man ner till vatten också
from there and you came down to water too
Readers. The failures of the readers consisted chiefly of not mentioning spatial relations. Readers also gave fewer locations than players.
No other systematic errors were found in the readers’ spatial descriptions.
6.2.5 Discussion
6.2.5.1 Description order
The order in which the locations were mentioned in the descriptions by the players as well as the readers did not match the order in
which they were visited for the most part. In contrast, Taylor and
Tversky (1992) found that the order in which participants drew
landmarks matched the order in which landmarks were presented in
a text describing a spatial layout. However, the many differences
between Taylor and Tversky’s study and the present one make a
comparison difficult. The length of the text, the number of landmarks, the time duration that the participants were allowed to
study the text, the instructions given to the participants, and the
fact that their participants drew a map as compared to giving description in speech are some potentially influencing factors. The
order in which the participants in the present study mentioned the
locations was not consistent with doing a scan over a spatial representation. Even though some descriptions matched visited/read order and spatial proximity, neither visited/read order nor spatial
proximity receives support as a general explanation for why participants mentioned the locations in the order they did. This suggests
that participants used some other strategy when recalling the locations than simply remember them in the order they were visited, or
scan a spatial mental representation.
Long-term memory representations of spatiality 165
In the players’ mentioning of the locations, there was a primacy
effect. The first three locations were given by all players. An explanation for this is that an initial obstacle was presented to the players
(how to get into the real estate office) and they spent time trying to
solve this puzzle, thereby elaborating these places during encoding
in memory.
6.2.5.2 Completeness, accuracy, consistency, integration
Completeness. Players as well as readers mentioned a fair number of
the locations from the game world, but players were more complete
in their descriptions.
In studies of learning from graphical maps, Taylor and Tversky
(1996) found that a mean of 94.6 percent of landmarks were mentioned by participants. Of course, this depends on the number of
landmarks to be learnt, but the Taylor and Tversky figure is higher
than the one obtained here for players and readers. The players
should have had plenty of opportunity to learn the locations. The
readers, on the other hand, may have made less effort to learn the
locations, so the low figure of the readers is more expected. An explanation for the low number can be the presence of a bias in the
completeness measure. It describes the spontaneous completeness
given by the participants, since they were not instructed to describe
the places from the game exhaustively, and no follow-up questions
were asked when they gave their descriptions. To some extent,
there is a chance element to the completeness measure used here.
Whether memory retrieval is triggered in order to further describe
the spatial layout in part depends on cues that happen to be available from, for example, the interviewer’s utterances or the participant’s associations. Thus, it is likely that the completeness measure
presents an unfairly low figure. If prompted, participants may have
given more locations and the completeness figure would go up.
However, there is no reason to believe that this would have affected
players and readers differently.
Nevertheless, there is a difference between players and readers
concerning completeness. What could be the reason for this? At
least three possible explanations suggest themselves. First, players
166 Participating in a story
were exposed to the locations a greater number of times than readers. This repetition effect may have increased the strength with
which the locations were encoded. Another explanation could be
that players were more interested and focused on the locations,
since they present an important part of playing Anchorhead—players
need to move between locations to explore and solve the game. In
this way, locations may have been elaborated in memory. The readers, in contrast, may have put less effort into thinking about the
locations, and the locations were thereby less elaborated. Finally,
consistent with the difference in completeness, players, but not
readers, may have formed a coherent spatial mental representation—
as a result of the navigational demand—which would make remembering locations easier.
Accuracy of spatial relations. Players provided highly accurate descriptions of the spatial relations, while readers did so less. Both
players and readers made mistakes, but players gave fewer inaccurate spatial relations.
The numbers are comparable to studies by Taylor and Tversky
(1992), where map drawings were made from learnt written route
descriptions. The means of four experiments showed an 82–90 percent accuracy, although the conditions were different from those in
the present study.
The difference in the accuracy of spatial relations between players and readers could be explained by the processes of repetition and
elaboration of single items. However, the difference between players and readers was even larger concerning spatial relations than
concerning locations. This strongly suggests that the difference is
due to the fact that players constructed coherent spatial mental representations.
Consistency. There were no occurrences of contradictions in spatial relations in the descriptions made by players, even though a
large number of spatial relations were mentioned. Readers gave very
few spatial relations, which made it unlikely that contradictions
could occur (the consistency measure does not reveal much in the
case of the readers).
Long-term memory representations of spatiality 167
The consistency of the players goes well with the idea that they
were using a spatial mental representation as a source for their descriptions.
Integration. Players spontaneously gave descriptions that consisted of integrated wholes. Participant 4 was an exception, describing six unconnected regions, but considering the difference eliciting
conditions, this is expected (he was not asked to give a description
of the spatial layout, but gave spatial description when answering
other questions).
Because the readers gave few or no spatial relations, their descriptions can be said to show a very low degree of integration, if the
measure could be meaningfully applied at all.
Taken together, these four measures support the conclusion that
players were indeed making their descriptions from spatial mental
representations, and the readers were not.
6.2.5.3 Hierarchical organisation
Players as well as readers expressed in their descriptions a hierarchical spatial organisation of the game world. However, this finding
does not warrant the conclusion that players and readers had hierarchical spatial mental representations in long-term memory. As
Brockmole and Wang (2002) point out, when using verbal descriptions as a measure of the spatial representation, it will proceed in an
ordered and hierarchical way, because of demands of the communicative situation. So, this could be attributed to the memory retrieval
and language production processes and is not necessarily a property
of the underlying mental representation. On the other hand, hierarchical organisation is something we would expect to find in participants’ descriptions, because hierarchical organisation is a fundamental property of spatial mental representations (Tversky, 2000). If no
signs of hierarchical organisation were to be found in the descriptions, we would be reluctant to conclude that the descriptions were
made from spatial mental representations.
168 Participating in a story
6.2.5.4 Errors
Two types of systematic errors were found in the players’ spatial
descriptions: a probable case of alignment bias, and confusion between the directions east and west.
As demonstrated by Tversky (1981), and many after her, it is
typical of spatial mental representations to be biased in a number of
ways. One type of error, alignment bias, seemed to be present in the
spatial descriptions made by the players. Given that it is indeed a
case of alignment bias, this suggests that the spatial description was
made from a spatial mental representation.
There were other errors in the spatial descriptions which suggest
a spatial mental representation. It was found that east and west were
confused more often than other pairs of directions. This suggests a
mapping of the terms east and west to a spatial organisation, because
people tend to more often confuse symmetrical axes than asymmetrical ones in spatial arrangements (Tversky, Lee, & Mainwaring,
1999). The explanation given by Tversky, Lee, and Mainwaring is
that ‘north-south are absolute directions, anchored at the poles,
whereas east-west are relative terms’ (p. 7). They also present findings that left and right are confused more than front/back or
head/feet. These findings more directly relate to an experience of
symmetry and asymmetry. Then, another explanation for the confusion of east and west—not mentioned in their article—would be
that they are spatially mapped onto right and left respectively. This
is the standard way of reading a map and may therefore influence
the spatial experience of the game world. Regardless of whether
east-west are inherently confusing, or whether they are confusing
because they are mapped to right–left, the errors strongly suggest an
underlying spatial mental representation.
The occurrences of mistaken directions, be it confusion between
east–west or other errors in directions, were associated with signs of
audible hesitation. Vowels were noticeably longer in cases of mistaken spatial relations than in cases of correct spatial relations. This
observation suggests that, to the extent participants were using spatial mental representations as a source for their descriptions, they
were at least partly constructing the spatial mental representation at
Long-term memory representations of spatiality 169
the time they gave the description. The reason is this: If participants
on the other hand did construct a spatial mental representation at
the time they played the game, the incorrect directions would be
part of the spatial mental representation in just the same way as the
correct ones. When retrieving directions from this spatial mental
representation, all directions—correct as well as incorrect ones—
would be treated equally. But instead, as was seen, many of the incorrect directions were reported in a different manner than the correct ones. Thus, at least part of constructing a spatial mental representation had to be performed at the time the descriptions were
made.
6.2.5.5 Why did the players construct spatial mental representations?
There are at least three possible reasons why players formed spatial
mental representations. First, it could be that having spatial mental
representations simply facilitates navigation in the story world. Second, closely connected to the first reason, it could be that not only
navigation, but also understanding of the participatory story, and
knowing which actions to perform, is facilitated by having a ‘mental map’ of the story world. Finally, as opposed to the other two
more functionalistic explanations, the spatial mental representation
may serve a purely aesthetic purpose in that participants enjoy the
participatory story more when they can ‘see’ the participatory story
world in their mind’s eye. It is not possible to rule out any of these
three explanations on the basis of the data in the present study. On
the other hand, there is little evidence to support any other conclusion than that the spatial mental models facilitate navigation. There
are no reasons to believe that actions other than navigation would
be facilitated by spatial mental representations. If the spatial mental
representations served a purely aesthetic purpose, readers of nonparticipatory stories would form them as well, but this does not
seem to be the case. Thus, it seems that players form spatial mental
representations because it helps them navigate in the game world.
170 Participating in a story
6.3
CONCLUSION
In summary, the present study of audience spatial cognition provides evidence that there is a difference in audience cognition between participatory stories and non-participatory stories. Participants who played the computer game Anchorhead supplied longer,
more elaborate verbal descriptions of spatiality of the game world
than participants who read the non-participatory version of Anchorhead. Players remembered spatial facts better. This was supported by more complete descriptions of locations and more accurate descriptions of spatial relations. It is suggested that the reason
for this improved memory performance is that players formed coherent spatial mental representations in long-term memory. This
was supported by the findings that their verbal descriptions of spatiality were highly consistent and presented an integrated spatial
whole. The forming of spatial mental representations by players
was to some degree supported by confusion of the symmetrical axis
east–west. Consistent with spatial mental representations, players’
verbal descriptions showed spatial hierarchical organisation.
Consequently, it seems that despite the often claimed fragmentation of interactive fiction, players were able to form a coherent spatial mental representation.
It could be argued that the non-participatory version of Anchorhead that was constructed for this study may be an atypical text
and not like an ordinary short story. It has an uncommon perspective (second-person pronouns) and may be less interesting than an
authentic short story. However, although the non-participatory
version acted as a point of comparison, the weight of evidence for
the construction of spatial mental representations in authentic reading situations of non-participatory stories can be found in other
studies, such as those by Zwaan and van Oostendorp (1993) and
Hakala (1999).
Nevertheless, the results of the present study give strong support
to the conclusion that players did construct spatial mental representations of the game world. The reason is probably that they did so
because they needed to. There is a need to move around—a naviga-
Long-term memory representations of spatiality 171
tional demand—in Anchorhead, as in almost all works of interactive
fiction. Thus, the hypothesis which states that players only use information locally in each situation in order to navigate can be refuted. As the navigational demand is present in most participatory
stories, it is likely that the audience constructs spatial mental representations in other kinds of participatory stories as well.
7
CHAPTER 7
Exploring memory qualities using the
reality monitoring framework
P
R O P O S I N G A F R A M E W O R K called reality monitoring (R M),
Johnson and colleagues (Johnson & Raye, 1981; Johnson et al.,
1988) have shown that memories of events that actually happened
differ in quality from memories of events that were merely imagined. Memories of real events generally contain more perceptual
detail, contextual embedding (such as time and place), and connections to other memories before and after the event. In contrast,
memories of imagined events generally contain more information
about thoughts and reflections from the time of the imagining.
Can the R M framework shed light on the qualities of memories
from a participatory story? Considering that the events that take
place in the participatory story are fictive, but at the same time
come about from actions carried out by the person experiencing the
story, the following questions present themselves: Are memories of
events from a participatory story qualitatively like memories of real
events, or like memories of imagined events, or do they present a
case in between these two positions? In other words, do they contain a high degree of perceptual details and contextual embedding
(like real events), or do they rather contain information on thoughts
and reflections (like imagined events), or are the kinds of information in memory in between these two? Using the R M framework,
this chapter explores the elicited interview data where people talk
about events from five sources (described in Chapter 5) in order to
answer these questions.
173
174 Participating in a story
7.1 REALITY MONITORING, DECEPTION
DETECTION, AND RATING OF VERBAL ACCOUNTS
First in this section, the R M framework is discussed. After that, earlier studies are investigated where R M has been applied to find differences in verbal statements of actual versus imagined events in the
area of deception detection—approaches which served as a model
for the present study of memory qualities.
7.1.1 Reality monitoring
Reality monitoring is a framework put forward by Johnson and
colleagues (Johnson & Raye, 1981; Johnson, Foley, Suengas, &
Raye, 1988) that addresses the question of how a person discriminates between memories that have an external (perceived) or an internal (imagined) origin. According to the theory, the origin of a
memory is not stored explicitly in memory, but is rather inferred
from various sources of information, among other things, the qualities of the memory. Memories of events that have been experienced,
that is, that have an external origin, contain more perceptual, spatial, temporal, emotional details. In contrast, memories of events
that have been imagined, that is, that have an internal origin, contain more elements of cognitive operations such as thoughts and
reasonings. When a person tries to remember whether a certain
memory corresponds to an actual experience or not, both conscious
as well as non-conscious processes are at work. Whether a memory
is attributed to an external or internal source depends in part on the
qualities contained in the memory. Although misclassifications occur, most of the time, these processes arrive at correct
classifications. To measure the degrees of various qualities of memories, Johnson et al. (1988) developed a memory characteristics questionnaire (M C Q), where participants rated their own memories.
The R M framework has not previously been applied to memories
from participatory stories, but has been useful in explaining findings
in areas such as post-event memory suggestion (Schooler, Gerhard,
& Loftus, 1986), dreams (Johnson, Kahan, & Raye, 1984), and deception detection (further discussed in detail below).
Memory qualities 175
Schooler, Gerhard, and Loftus (1986) studied the qualities of
memories of real versus suggested details in a scene as revealed
through written descriptions by participants. The study seems to be
the first to take the step from studying people’s ratings of their own
memory qualities to studying the presence of various types of information in external, verbal descriptions of memories. In the study,
judges rated the written descriptions on sensory, geographic (i.e.,
spatial), functional, and cognitive-processing information, as well as
verbal hedges (e.g., ‘I think…’, ‘I believe…’). The real-memory descriptions had more sensory and geographic information, while the
suggested-memory descriptions had more functional and cognitive
processing information as well as more verbal hedges. Thus, it
seems reasonable to make a connection between, on the one hand,
memories of external events and non-suggested memory, and on the
other hand, memories of internal events and suggested memory.
Next, one particular application of the R M framework is considered which includes external ratings of memory qualities, offering
promising methods for the present study: the detection of deception.
7.1.2 Deception detection using verbal accounts
The R M framework has been applied to the problem of discriminating truth from lies by researchers working in the field of deception
detection within forensic psychology. The general procedure is,
first, to obtain spoken accounts from people who have witnessed
actual events or have imagined events. These accounts are then transcribed and subsequently rated by judges who score the contents on
a set of R M criteria. In contrast to the M C Q , where people rate their
own memories, in deception detection research, judges rate other
people’s accounts. These characteristics make R M applied to deception detection suitable for the present study where data also was
collected in the form of spoken accounts. A general assumption in
the use of R M in deception detection is that false testimonies can be
likened to memories of internal events, and that true testimonies
can be likened to memories of external events.
176 Participating in a story
Alonso-Quecuty (1992) was the first to test whether the R M
framework could be used to distinguish between true and false testimonies, and also investigated the influencing effect of a time delay
(10 min) between experience and testimony. Participants were divided into two groups. In Group 1, participants watched a videotape of a person performing a criminal act. Half of the participants
were instructed first to tell what they saw on the videotape, and
then give an account that made the person not guilty of having performed the criminal act. The other half gave the accounts in the
reverse order. Participants in Group 2 received the same treatment
as those in Group 1 with the insertion of a 10-minute delay before
giving the accounts. Five measures were taken on the accounts:
number of words, number of pauses, and ratings of sensorial, contextual, and subjective idiosyncratic information. In the paper, no
details are given about the last measure, but presumably, it coincides
with the R M property cognitive operations, that is, reasoning, inferring, etc. Results showed that false testimonies were longer and contained more pauses when there was a delay. The immediate true
accounts had more sensorial and contextual information, while the
immediate false accounts had more subjective idiosyncratic information. In the delayed condition, only the amount of idiosyncratic
information differed, being higher in the false accounts. Thus, the
study appears to show that the R M framework can be used for deception detection, at least when the testimonies are made without
delay.
In a similar study, Hernández-Fernaud and Alonso-Quecuty
(1997) found more contextual and sensorial information in true
statements, but no difference concerning cognitive operations in
true compared to false statements, as would be expected from the
R M framework.
Vrij, Edward, Roberts, and Bull (2000) studied, among other
measures, the presence of R M criteria in true and false accounts.
Results concerning R M criteria showed that perceptual (vision,
sound), spatial, and temporal information were present to a higher
degree in true than in false accounts. Contrary to R M predictions,
cognitive operations were also more present in true accounts. Vrij,
Memory qualities 177
et al. (2000) explain this finding by noting that cognitive operations
are one way of facilitating memory for experienced events, by using
the context to reason about the likelihood of actually having experienced the event (e.g., a person may believe she has driven a car fast
in Germany, because it is a likely consequence from remembering
having used the motorway).
In a recent study by Vrij, Akehurst, Soukara, and Bull (2004), results were consistent with R M predictions. Visual, auditory, spatial,
and temporal details were present to a higher degree in true than in
false statements. Cognitive operations were present to a higher degree in false statements.
Another application of R M in deception detection was made by
Sporer (1997). Sporer noted that the methods of content-based
analysis of accounts in criminal contexts lacked psychological theory, and suggested that the R M framework can fill this role. In
Sporer’s study, participants were first instructed to tell about real
and invented events (either immediately or with a short delay)
while being video recorded (the terms ‘lie’ and ‘truth’ were deliberately not used in the instructions). These video recordings were
transcribed and rated by judges on the degree of various kinds of
information related to the R M framework. The criteria used by
Sporer were a reduced version of the M C Q , where several dimensions had been collapsed into eight scales, based on factor analyses
and theoretical assumptions of the R M framework. Sporer’s criteria
were adapted for judgement of accounts by external observers.
Judges rated the criteria on a three-graded scale. The results showed
a higher presence of spatial, time, and emotional information, but
only when there was a short delay before giving the account. Averaged across the immediate and delay conditions, the results showed
a higher presence of emotions and realism in the true accounts. The
R M criteria could successfully discriminate true stories from false
stories (71.3 percent correct, which is considered a high figure
within the field of deception detection).
178 Participating in a story
In summary, the application of the R M framework in deception
detection has proven partly successful. The presence of perceptual
and contextual details is most often found to a higher degree in accounts of real, experienced events. However, the evidence that cognitive operations are present to a higher degree in accounts of imagined events is mixed—most studies show more cognitive operations
in false statements, but some studies show no differences, and some
studies even show a higher degree of cognitive operations in true
statements, contrary to predictions from the R M framework.
Because of the success of Sporer’s (1997) study as well as the relatively detailed documentation of it, its method was used as a basis
for the present study. In applying Sporer’s method in the present
study, a parallel is assumed between, on the one hand, external and
factual events, and on the other hand, between internal and fictional
events. The assumptions of the present study are put in relation to
assumptions of deception detection in Table 7.1. However, in making the connection between fictional and imagined events, it is important to add that it is not claimed that fiction is the same as lies
(see Chapter 3 for a discussion of fiction).
Table 7.1. Assumptions regarding the status of events based on the reality monitoring framework in deception detection and the present study.
Reality monitoring
Detection deception
The present study
Memories with
external origin
Truth
Factual
Memories with
internal origin
Lie
Fictional
Memory qualities 179
7.2
REALITY MONITORING CRITERIA
The rating criteria in the present study consisted of nine elements,
described below. Criteria 1–7 are those typically present to a higher
degree in memories of real, external events, while criterion 8 typically is present to a higher degree in memories of imagined events.
The first six criteria were taken from Sporer (1997). The criterion of
Degree of details was added because this quality separates external
and internal memories according to the R M framework (Johnson &
Raye, 1981). Further, Sporer’s criterion Cognitive operations was
divided in two criteria, Cognitive operations at the time of the event,
and Cognitive operations at the time of retelling. This division was
made in order to sort out cognitive operations tied to the memory
itself, for example, ‘I thought that he was going to run after me’
(which according to the R M framework are associated particularly
with internal memories), in contrast to cognitive operations occurring as part of the telling, for example, ‘When I think about it now,
it seems that what happened was…’. Schooler, Gerhard, and Loftus
(1986) use the term verbal hedges to refer to what is here called Cognitive operations at the time of retelling. Without this division, there
was a risk that raters would collapse both criteria, when only criterion 8 is of interest within the R M framework. Criterion 9 is related
to the conversational level of the discourse, and does not correspond to qualities of the memories themselves. A characteristic is
that for Cognitive operations at the time of retelling, the verb is virtually always in the present tense, while for Cognitive operations at the
time of the event, the verb is usually in the past tense.
The requirements for judging whether R M criteria are present are
as follows:
1.
Clarity. This refers to the clarity and vividness of the statements. This criterion is present if the report is clear, sharp,
and vivid (instead of dim and vague).
2.
Perceptual information. This criterion is present if the
statement includes sensorial experiences such as sounds
(e.g., ‘He really shouted at me’), smells (e.g., ‘It had a smell
180 Participating in a story
of rotten fish’), tastes (e.g., ‘The chips were very salty’),
physical sensations (e.g., ‘It really hurt’), and visual details
(e.g., ‘I saw the nurse entering the ward’).
3.
Spatial information. This criterion is present if the statement includes information about locations (e.g., ‘It was in a
park’) or the spatial arrangement of people and/or objects
(e.g., ‘The man was sitting left of his wife’ or ‘The lamp
was partially hidden behind the curtains’).
4.
Temporal information. This criterion is present if the
statement includes information about when the event happened (e.g., ‘It was really in the morning’) or explicitly describes a sequence of events (e.g., ‘When he heard all that
noise, the visitor became nervous and left’, ‘As soon as the
guy entered the pub, the girl started smiling’).
5.
Affect. This criterion is present if information is included
about how the participant felt during the event (e.g., ‘I was
very scared’).
6.
Reconstructability of the story. This criterion is present if it
is possible to reconstruct the event on the basis of the information given.
7.
Degree of details. This criterion is present if the account has
a high degree of detail (e.g., ‘The door-knob was made of
brass’, ‘He opened the envelope eagerly’).
8.
Cognitive operations at the time of the event. This criterion is
present if there are descriptions of conclusions, reasoning,
or thinking that took place at the time of the event or
events (e.g., ‘Then I decided to…’, ‘Her reactions gave me
the impression that she was upset’).
9.
Cognitive operations at the time of the telling. This criterion
is present if there are descriptions of conclusions, reasoning
or thinking occurring at the time of the telling (e.g., ‘I
think this is what happened…’, ‘As I remember it…’, ‘It
strikes me now that…’).
Memory qualities 181
(The description of the first six of Sporer’s criteria is quoted from
Vrij (2000), p. 160; some parts of another rating system are left out.)
Sporer (1997) used an additional R M criterion, Realism, but this
was considered irrelevant in the present study and was removed.
The reason was twofold: The content of the stimulus stories was
often given to the participants who made the account, giving them
little chance to influence it and therefore the degree of realism of
their tellings. Further, the source of each account was often explicitly stated in the accounts, for example, ‘I remember when I was ten
years old…’, ‘The game started with…’, and ‘Sherlock Holmes deduced that the criminal must have…’. This makes the issue of realism meaningless in the present study, since judges could easily tell
from what source the account was made. In contrast, Sporer used
the criterion because his judges did not know from which condition
the accounts came.
7.3
HYPOTHESES
As described in Chapter 5, the participants’ accounts were from five
sources:
•
Game: Events from a computer game which the participants played.
•
Story: Events from a short story which the participants
read.
•
Personal: Events that took place earlier in the participant’s
life.
•
Tasks: Events from three tasks carried out in a laboratory
setting.
•
Non-participatory: Events from a special, adapted nonparticipatory version of the computer game, printed on
sheets of paper, which the participants read.
182 Participating in a story
According to Johnson and Raye (1981), Johnson et al. (1988), and
Sporer (1997), the R M criteria 1 to 7 (described above) should be
present to a higher degree in external, real events. The R M criterion
8 should be present to a higher degree in internal, imagined events
(as discussed, criterion 9 does not concern R M issues, and is therefore not part of the hypotheses). The hypotheses were that memories of events read about in a short story (Story) share qualities of
imagined events, and that memories of events from a computer
game (Game) share some of the qualities of memories of actual
events as well as some qualities of imagined events. Further, it was
hypothesised that memories of events from personal experience
(Personal) share qualities with memories of events from the laboratory tasks (Tasks), and that memories of events read about in a short
story (Story) share qualities of memories of events from the nonparticipatory version of the computer game (Non-participatory).
Thus, the hypotheses regarding the rating scores on the R M criteria
in the present study were as follows:
RM
criteria 1–7 (perceptual, spatial, time, affect, etc.):
Personal = Tasks > Game > Story = Non-participatory
RM
criterion 8 (cognitive operations):
Personal = Tasks < Game < Story = Non-participatory
Memory qualities 183
7.4
METHOD
All material and instructions are presented here translated into English from Swedish, which was the language used for conducting the
study.
7.4.1 Data processing
7.4.1.1 Accounts
From the transcriptions of the interview data from the eight participants (see Chapter 5), the accounts of events therein were extracted. Deciding what constitutes a single account or several accounts is difficult to do formally and consistently (for a discussion
of this problem in relation to everyday narrative, see Labov &
Waletzky, 1967). In the present study, the start of accounts was chosen according to several criteria: content, the occurrence of a long
pause or an inhalation, or typical occurrences such as ‘Let me tell
you about when I…’. The end of accounts was chosen according to
content or typical summing-up phrases. The beginning and end of
accounts were also signalled by non-feedback turn-taking. All nonword elements from the account transcriptions were removed. All
conversational turn-taking was removed, leaving only what appeared to be an account given in monologue. If accounts were less
than 50 words and they belonged to a more general account (if the
theme was similar, i.e., it was a partial account), it was merged with
the account following it. Accounts longer than 500 words were cut
approximately at that point (letting them end as naturally as possible). This cut off two accounts, otherwise being 2,432 and 1,002
words. The upper length of 500 words was chosen as a practical
limit for the rating procedure. Any potential qualities were considered to be well present within the first 500 words. An example of an
extracted account, in the format presented to the judges, is shown in
Table 7.2 (here with an added English translation).
184 Participating in a story
Table 7.2. Example of an extracted account used in the study.
Swedish
English translation
det visar sig att man är en kvinna som är
gift med en man som heter michael och
som har fått reda på att han har släktingar i nån nån liten stad där både fiskeindustrin och pappersindustrin har gått
omkull då tydligen får man reda på sen
eh man flyttar dit och ska hämta nycklarna på ett fastighetskontor men där är
det stängt och låst å man kan ta sig in
genom ett fönster på baksidan och där
inser man att eh man kan lyssna på en
telefonsvarare å få reda på efternamnet
på familjen sen går man å letar efter
familjen i file room där hittar man nycklar men man hittar också att alla papper
är borta sen ska man hitta sin eh man då
på universitetet eh ta mannen å gå till
huset å sen ska man gå och lägga sig sen
tar det slut
it turns out that you are a woman who is
married to a man named michael who
has learnt that he has relatives in some
some small town where both the fishing
industry and the paper industry have
been shut down evidently you learn later
eh you move there and is supposed to
get the keys at a real estate office but it is
closed and locked and you can get in
through a window at the back and there
you realise that you can listen to a telephone answering machine and learn the
name of the family then you go to search
for the family in the file room there you
finds keys but you also find that all the
papers are gone then you’re supposed to
find your eh husband at the university
eh get the husband and then go to the
house and then you’re supposed to go to
bed and then it ends
A total of 32 accounts were extracted. The number of accounts per
source varied between 4 and 9. The accounts varied considerably in
length. A one-way analysis of variance (A N O V A) showed that the
mean lengths of the stories from the five sources differed, F (4,27) =
3.365, p = .023. However, given that the ratings were made globally
for each account, they should not be sensitive to differences in
length, even if these are systematic.
7.4.1.2 Rating questionnaire
The rating questionnaire consisted of general information about the
study, descriptions of the nine R M criteria together with examples,
description of the rating scale, some background questions on age,
gender, and first language of the judge, followed by three practice
accounts and the 32 accounts (in randomised order for each judge).
Memory qualities 185
After each account, the nine criteria was listed followed by a threegraded rating scale, taken from Sporer (1997) (0 = not present, 1 =
some indication present, 2 = clearly present). The three-graded rating scale was used instead of the seven-graded scale of the M C Q
(Johnson et al., 1988). The rationale was that a seven-graded scale is
suitable when the rater has rich information from which to rate, as
is the case with the M C Q when rating one’s own phenomenological
qualities of memories. When rating the sparse information in brief
verbal accounts, very little information is available on which to base
a rating, making a two- or three-graded scale more suitable.
7.4.2 Rating procedure
Two judges, one male and one female, both with Swedish as their
first language, rated each of the 32 accounts on the nine R M criteria.
The judges were both students with some experience of reading
transcribed speech, but unaware of the research hypotheses. Prior to
rating, the judges were briefly trained in using the R M criteria. After
an introduction to the meaning of each R M criterion, the first account was rated together with the experimenter and the ratings discussed. The second and third accounts were rated separately, but
discussed together. After this training, the judges were considered to
have reached a sufficient level of understanding of the R M criteria,
and they rated the 32 accounts separately.
7.5
RESULTS
Results are here presented first for inter-rater agreement, then for
the rating scores of the R M criteria.
7.5.1 Inter-rater agreement
The inter-rater agreement was calculated with percent agreement
and Pearson product moment correlation, shown in Table 7.3. The
percent agreement varied between 46.9 and 84.4 with a mean of 59
186 Participating in a story
percent for all criteria. Inter-rater correlation was significant at the
p < .01 level for Clarity, Perceptual information, Spatial information, Time information, and Affect, while Degree of details and Cognitive operations at the time of the event were significant at p < .05.
Table 7.3. Inter-rater agreement of the R M criteria ratings of the two judges.
% agreement
Pearson r
1. Clarity
68.75
.467**
2. Perceptual information
59.375
.564**
3. Spatial information
68.75
.714**
4. Time information
53.125
.507**
5. Affect
84.375
.702**
6. Reconstructability
53.125
.103
7. Details
46.875
.444*
8. Cognitive oper. at event
50
.441*
9. Cognitive oper. at telling
46.875
.082
RM
criteria
*p < .05; **p < .01.
7.5.2
RM
criteria
The mean of the ratings of the two judges were used in the subsequent analyses—values ranging from 0.00 to 2.00.
Descriptive analyses revealed that the fifth criterion, Affect, was
hardly ever coded to be present in the accounts (M = 0.11), and was
therefore excluded from further statistical analyses. There was considerable variation within the ratings of the other criteria, and
thereby presumably in their discriminative value.
A N O V A s investigating effects of identity and gender of the participant who made the account on ratings showed no significant
effects.
Length of accounts correlated with Time information (R M criterion 4), r = .426, p = .015, and with Cognitive operations at the time
of the telling (R M criterion 9), r = .402, p = .023.
Memory qualities 187
The rating means for accounts of the five sources, for each R M
criterion, can be seen in Table 7.4. Table 7.5 shows the collapsed
means of R M criteria 1–4 and 6–7, as well as R M criterion 8 (as discussed, criterion 5 was discarded from analyses because of a floor
effect).
The hypothesis that ratings for criteria 1–7 would be highest for
Personal and Tasks, with Game in between, and Story and Nonparticipatory lowest, was not supported by the data (one-way
A N O V A with source and means of R M criteria 1–4, 6–7, F (4,27) =
0.146, n.s.) However, comparison of individual R M criteria against
source revealed a significant difference for Spatial information (R M
criterion 3) (one-way A N O V A with source and R M criterion 3,
F (4,27) = 7.264, p < .001). Multiple post hoc comparisons using
Tukey H S D showed that for Tasks, Spatial information was rated
lower than for Story (md = 0.9167, p = .005), Non-participatory
(md = 1.1250, p = .007), and Game (md = –1.2500, p = .001).
The hypothesis that ratings for Cognitive operations at the time of
the event (criterion 8) would be highest for Non-participatory and
Story, with Game in between, and Personal and Tasks lowest, was
not supported by the data. No difference across sources was found
(one-way A N O V A with source and R M criterion 8 yielded F (4,27) =
0.658, n.s.)
Table 7.4. Means of R M criteria ratings as a function of account source.
criteria
1. Clarity
2. Perceptual information
3. Spatial information
4. Time information
6. Reconstructability
7. Details
8. Cognitive oper. at event
9. Cognitive oper. at telling
RM
Personal Tasks
1.50
1.75
0.60
1.06
0.80
0.25
1.00
1.13
1.40
1.63
0.50
0.69
0.60
1.13
0.50
0.25
Game
1.33
0.42
1.50
1.25
1.50
0.50
0.92
0.17
Story
1.44
0.78
1.17
0.83
1.33
0.78
0.67
0.44
Non-p.
1.38
0.13
1.38
1.38
1.50
0.88
0.75
0.63
Scales range from 0.00 to 2.00. Criterion 5 was excluded from analyses because of a
floor effect.
188 Participating in a story
Table 7.5. Means of R M criteria ratings for the five sources on criteria 1–4, 6–7 and
criterion 8.
Source
Criteria 1–4, 6–7*
Criterion 8
Personal
0.97
0.60
Tasks
1.08
1.13
Game
1.08
0.92
Story
1.06
0.67
Non-participatory
1.10
0.75
Note: Scales range from 0.00 to 2.00. *Criterion 5 was excluded from analyses because of a floor effect.
Correlations between R M criteria were calculated in order to see if
they represented independent factors, or if some criterion was redundant with respect to others. Calculation of Pearson r, N = 32,
two-tailed, revealed five significant correlations at the .01 level, as
shown in Table 7.6. Correlation between R M criteria is further discussed below.
Table 7.6. Significant correlations among R M criteria.
RM
criteria
RM
criteria
r
p
1. Clarity
6. Reconstructability
.456
.009
1. Clarity
7. Details
.551
.001
1. Clarity
8. Cognitive oper. at event
.479
.006
4. Time information
6. Reconstructability
.523
.002
4. Time information
8. Cognitive oper. at event
.536
.002
Note: Correlations were calculated with Pearson r, N = 32, two-tailed, p < .01. p is
the level at which the correlations are significant.
Memory qualities 189
7.6
DISCUSSION
The goal of the present study was to see whether the R M framework could shed light on qualities of memories from a participatory
story by applying external ratings of R M criteria to verbal accounts.
The results showed no support for the hypothesis concerning the
higher presence of perceptual, contextual, and emotional information (criteria 1–7) in events from personal experience and laboratory
tasks (real, actual events), nor any support for the hypothesis concerning higher presence of cognitive operations (criterion 8) in
events from a short story and a non-participatory printed version of
interactive fiction. The only criterion that differed across sources
was Spatial information (criterion 3), which was present to a lower
degree in events from laboratory tasks than in events from interactive fiction, the short story, and the non-participatory version.
One reason for caution is the quasi-experimental nature of the
study. The sample size was small (N = 32), the influence of individual variation was only partially balanced, and the content varied
between the conditions (see Chapter 5 for more details about the
method). However, if people discriminate memories in the way
suggested by the R M framework, effects could be visible even given
these methodological shortcomings.
7.6.1 Inter-rater agreement
The criteria Reconstructability had a low inter-rater correlation (r =
.103) and a rather low inter-rater agreement (53.13 percent). However, this low value was because of the three-point scale used. If the
criteria are recoded by combining ratings of 1 and 2 to a single value
(thus dichotomising the scale), the inter-rater agreement rises to
93.75 percent. Based on this, it is reasonable that Reconstructability
was part of the analysis. The criterion Cognitive operations at the
time of telling also had a low inter-rater correlation (r = .082) and a
low inter-rater agreement (46.88 percent), but as this criterion was
not used in the analysis, it is of no concern in the present study.
The inter-rater agreement between 46.9 and 84.4 percent with a
mean of 59 percent is comparable to that of Sporer (1997), where
190 Participating in a story
the percentage inter-rater agreement of the two raters varied between 36.3 and 67.5 percent with a mean of 52.5 percent. Inter-rater
correlation was significant at the p < .01 level for five criteria and
significant at p < .05 for two criteria. This is also comparable to
Sporer’s inter-rater correlations that were significant at the p < .01
level for four criteria and at the p < .05 level for one criterion. Although the raters in the present study received only little training,
their ratings agreed as much as or more than the extensively trained
raters used in Sporer’s study.
7.6.2 Presence of criteria 1–7 in memories of events
The hypothesis that ratings of criteria 1–7 would be highest for Personal and Tasks, with Game in between, and Story and Nonparticipatory lowest, was not supported by the data. Thus, it appears that in accounts, the qualities associated with external events
are not present to a higher degree in interactive fiction than in a
short story, or even in a manipulated non-participatory version of
interactive fiction, otherwise identical. Neither is the presence of
such qualities lower in interactive fiction than in accounts of personal experience or laboratory tasks, which would be expected considering the R M framework.
The only R M criterion that differed across sources was Spatial information. Contrary to expectation, the accounts of events from the
laboratory tasks contained less spatial information than all other
sources except events from personal experience. In contrast, Sporer
(1997) obtained higher ratings for true accounts for Spatial information (when accounts were made when the speaker was given extra
preparation time)—similar findings were made by HernándezFernaud and Alonso-Quecuty (1997), Vrij, et al. (2000), and Vrij, et
al. (2004). Also, Alonso-Quecuty (1992) found more contextual information (of which spatial information was a part) in immediate
true accounts. Similarly, Schooler et al. (1986) found more geographic information in non-suggested memories. The explanation
for the results in the present study is probably that the events from
the various sources differed in content. The laboratory tasks did not
Memory qualities 191
involve much spatial aspects, at least not beyond the space around
the body. Thus, there may have been little spatial information to
report. In contrast, the other sources contained events of spatial
movement on a larger scale.
The criterion Perceptual information showed no difference across
the five conditions. In contrast, Alonso-Quecuty (1992) found more
sensorial information in immediate true accounts, HernándezFernaud and Alonso-Quecuty (1997) found more sensorial information in true statements, Vrij et al. (2000) and Vrij et al. (2004) found
more perceptual information (both vision and sound) in true accounts, and Schooler et al. (1986) found more sensory information
in real (non-suggested) memory. This result in the present study is
surprising, since the two conditions of personal experience and
laboratory tasks should present typical cases of real, external events,
while the short story and the non-participatory version of the computer game should present the opposite—events that did not have
any external source.
In Sporer’s (1997) study, there were additional differences in R M
criteria between invented and true accounts. Time information and
Emotions differed when the accounts were made with extra preparation time for the speaker. Vrij et al. (2000) and Vrij et al. (2004) also
found more temporal information in true accounts. For the criteria
Emotions and Realism (the latter was not included in the present
study) Sporer (1997) obtained higher overall means for true accounts, regardless of preparation time. In the present study, Emotions received very few ratings and was as a result excluded from
further analysis. Perhaps a reason for the difference concerning
emotional content is that in Sporer’s study, the participants were
instructed to give accounts which included personally important
events. It is conceivable that the requirement of importance led to
accounts that were emotional. Although the present study contained
the possibility for emotional events (participants could choose personally experienced events, and participants could relate emotionally to the laboratory tasks—e.g., by being frustrated or bored by
them), there was no explicit instruction to give accounts of emotional or personally important events. Turning to the criterion
192 Participating in a story
Time information, the difference found by Sporer (1997), Vrij et al.
(2000), and Vrij et al. (2004) was not seen in the present study. The
criterion should be able to vary freely in the accounts from the
various sources in the present study, and nothing was found that
could explain the difference in this criterion compared to the results
the other studies mentioned.
The lack of difference between the personal experience condition
and the other conditions can partly be explained by the effect of
time. Johnson et al. (1988) obtained results that showed that actual
and imagined events become more alike as time passes. However,
the laboratory tasks condition is not influenced by time since it was
performed recent to giving the accounts.
7.6.3 The presence of cognitive operations
The hypothesis that Cognitive operations at the time of the event (R M
criterion 8) would be highest for Non-participatory and Story, with
Game in between, and Personal and Tasks lowest, was not supported by the data. Cognitive operations from the time of the event
are not present to a higher degree in accounts about events from
interactive fiction compared to accounts about personal experience
or laboratory tasks. Likewise, accounts about interactive fiction
events do not appear to contain cognitive operations to higher degree than accounts about personal experience or laboratory tasks.
This finding is consistent with that of Sporer (1997), where his
criteria Cognitive operations did not differ between invented and real
accounts, but inconsistent with Alonso-Quecuty’s (1992) results
that idiosyncratic information is present to a higher degree in false
testimonies (both immediate and delayed), and with Schooler et al.’s
(1986) results that there are more cognitive operations in descriptions of suggested memories, as well as with Vrij et al. (2000), who
found more cognitive operations in true than false testimonies.
The differences across various studies in results regarding the
presence of cognitive operations in verbal descriptions suggest that
there may be a problem of reliability concerning this criterion.
Some studies show the lowest inter-rater agreement on this criterion
Memory qualities 193
(Vrij et al., 2004; Strömwall, Bengtsson, Leander, & Granhag, 2004),
or next-to-lowest (Vrij et al., 2000). Vrij, Evans, Akehurst, and
Mann (2004) even failed to find a reliable inter-rater agreement score
for cognitive operations, and discuss that it may be difficult for raters to grasp the meaning of this criterion. For many studies, the
operationalisation of this criterion is not documented, making it
difficult to know whether the same characteristic is being rated
across studies. As done in the present study, one suggestion is to
separate two kinds of cognitive operations: those tied to experiencing the event, and those tied to reporting about the event. The former is usually given with the verb in the past tense. The latter, being on the here-and-now level of the discourse, virtually always includes verbs in the present tense.
Regarding the ratings of Cognitive operations at the time of the
event in the present study, a possible explanation of the results appears be that there may be an influencing factor other than the
source from which the events were taken. Looking at Table 7.5, it
appears that the ranking of low to high degree of information about
cognitive operations could be explained by the amount of focus on
problem-solving in each situation. A low focus on problem-solving
(personal experience) would give a low rating on the presence of
cognitive operations in the accounts, and a situation in which the
focus is on problem-solving (laboratory tasks) would result in an
account which is ridden with talk about cognitive operations. However, this explanation was not supported by the data, since there
was no significant difference in ratings between the sources.
7.6.4 Correlations of R M criteria
Analysis of correlations between R M criteria showed that several
criteria were correlated. One may consider that the correlations
were present because they were all correlated with the length of the
account. That is, longer accounts would mean more chance of any
criterion being present. However, the only criterion correlated with
length was Time information (criterion 4).
194 Participating in a story
The correlational analysis indeed showed that some criteria appear to be redundant. While some correlations do not seem to capture any structural properties of the R M criteria (e.g., Clarity and
Cognitive operations at the time of the event), the correlation of Reconstructability (criterion 6) with Clarity (criterion 1) as well as with
Time information (criterion 4) seems to suggest that these are three
aspects of the same underlying property. If an account contains time
information and is clear, it is also easier to reconstruct the whole
event. Thus, on the basis of these correlations, the criterion Reconstructability can be viewed as consisting of the more specific criteria
Clarity and Time information, which would make the Reconstructability criterion redundant in similar future studies.
7.6.5 Validity of the method
There are several possibilities related to the method used, explaining
why no support was found for the hypotheses concerning differences in qualities of memories in the present study.
First, it might be that differences in qualities are present in
memories, but that they are not transferred to language. One possible influence could be the situation in which the verbal accounts
were given. Participants might have adapted their accounts according to any number of expectations (whether controlled by the experimenter or not). This is not likely, because great care was taken
that the situation should be as similar across sources and as natural
as possible (see the extensive discussion of this point in Chapter 5).
However, there may still be some general reason why thought
qualities are not put into words. The original formulation of R M
was arrived at by self-rating of memory qualities. In contrast, the
present study uses ratings by external observers, which adds another
layer on top of the memories. However, Sporer (1997), AlonsoQuecuty (1992), and Schooler et al. (1986) have demonstrated that
qualities as described by the R M framework can be transferred from
memories to verbal accounts. In Sporer’s study, the invented accounts were not remembered by his participants. As a consequence
of his experimental procedure, the invented accounts were rather
Memory qualities 195
created in the experimental situation. However, the true accounts
were made by instructing the participants to tell about events that
had happened to them personally. Thus, qualities of memories of
real events were present in the verbal accounts.
Second, the reason for lack of support of the hypothesis in the
present study might be that the differences in qualities actually are
present in the accounts, but that the rating procedure was not sensitive enough to differentiate between the sources. However, the satisfactory results of the inter-rater analysis make this explanation
implausible. Also, the success of applying R M ratings on actual and
invented accounts, as was done in Sporer’s (1997) study, indicates
that the rating procedure has something going for it.
In summary, the ratings in the present study should reflect actual
qualities in memories of events, because these are transferred from
memory to language and the rating procedure should be able to detect them.
What is left, then, is the explanation that there are no qualitative
differences in memories of events from the five sources studied, at
least not those as isolated from the R M framework by Sporer (1997).
7.6.6 Alternative interpretations
How may the result that no differences were found between the five
conditions, as predicted by the R M framework, be explained?
7.6.6.1 A participatory story as similar to real personal experience
Could it be that the participatory story created as vivid and ‘external-like’ memories (in the R M framework) as real personal experience? This conclusion seems to be supported by the data, since there
was no difference in the ratings of the R M criteria of the Game and
Personal conditions. However, if this is so, then one is forced to
conclude that also the short story and the non-participatory version
created these kinds of vivid memories. Since this leads to a situation
in which nothing is explained, the initial interpretation that the
participatory story and real personal experience produce the same
kind of vivid memories was rejected.
196 Participating in a story
7.6.6.2 Factual versus fictional events
Is the R M framework really about how people distinguish memories
of actual events from memories of events that did not happen? Or is
R M about distinguishing perceptually-based (i.e., external) memories
from non-perceptually based (i.e., internal) memories. In other words,
can the parallel between external/real and internal/fictional be sustained? It seems that the R M framework makes the tacit assumption
that real events are always external and non-real events are always
internal. However, factual events can be both external (e.g., seeing
someone driving a car) and internal (e.g., reading the news). Fictional events can be both external (e.g., watching a play) and internal (e.g., reading a fictional novel). Thus, it seems that the R M
framework may not address the real/fictional distinction at all (see
Table 7.7).
A task for future research would be to tease apart the factual/fictional status of events from the R M framework’s external/internal status. An experiment could present the four conditions shown in Table 7.7 and afterwards let participants rate their
memories using Johnson et al.’s (1988) M C Q . Of course, a challenge
for such a study would be to control for differences in contents between the four conditions. An interesting question for future research would be to locate the position of memories of a participatory story in such a scheme.
A conclusion from the present study is that not only could an
R M rating not shed light on the status of memory qualities of interactive fiction, but nor could it reveal differences between accounts
of events from a short story as compared to real life personal experience. This has implications for psychological research on reading comprehension and the reading experience, particularly for the
reading of fiction. Using the R M framework, or at least Sporer’s
(1997) reduced set of criteria and ratings by external judges, in research on experiential qualities of reading would not be a promising
approach. The same may also hold for studies of film comprehension.
Memory qualities 197
Table 7.7. Factual, fictional, external, and internal events
Origin of memory
Fictional status of
event in memory
Factual
Fictional
External
Internal
RM
Addressed by the
framework
Not addressed by the
R M framework
Not addressed by the
framework
Addressed by the
R M framework
RM
7.7
CONCLUSION
Taken together, the results of this study suggest that qualities of
memories of events from sources such as a short story, a computer
game, and personal experience, do not differ along the R M dimensions. This goes counter to predictions from the R M framework,
under the assumption that memories of internal events are like
memories of fiction (e.g., things read in a short story) and memories
of external events are like memories of personal experience. If the
results are not influenced by methodological limitations, then the
explanation may be that the R M framework is simply not a suitable
candidate for an explanation of the difference between memories of
events with varying degree of fictionality. Teasing apart the external/internal distinction from the factual/fictional distinction is then
a task for future research. What the results imply for cognition and
the concept of participatory stories is discussed in Chapter 9.
8
CHAPTER 8
Perspective on actions and events
Tdo people adopt on actions and events from participatory stoHE QUESTION
of concern in this chapter is: What perspective
ries? The term perspective has a variety of meanings, but is here constrained to mean how a speaker mentally positions herself, both
spatially and regarding psychological distance, in relation to some
remembered event or action. Thus, perspective is here defined as a
cognitive construct.
As discussed in Chapter 4, at the heart of participatory stories is
the concept of agency. The audience carries out actions when interacting with a participatory story. How is experiencing a story different when you carry out actions in it compared to when you do
not (as is the case in non-participatory stories)? And how is it different to carry out actions in a participatory story compared to carrying out actions in the real world? How does the audience view
agency in participatory stories? Who is carrying out actions in a participatory story? Is it the audience or the character/agent in the
story? These questions connect to the general question of differences in audience cognition when comparing participatory stories
and non-participatory stories. More specifically, the questions concern what, if anything, is different about how events are cognitively
processed between participatory stories and non-participatory stories. Thus, perspective may be one such difference. When remembering events from one’s own personal experience, the perspective on
the memory is that one participated in the event. When talking
about agency in such situations, one would use the pronoun I. In
contrast, remembering events from a fictional text, such as a novel,
199
200 Participating in a story
would lead to a different perspective—that of an observer, looking
at the event from the outside. Here, when talking about agency, one
would use pronouns such as he or she, or nouns such as woman.
Considering that participatory stories contain the elements of fiction as well as action, the question becomes what perspective people
adopt on actions and events from participatory stories.
In her studies of Swedish children’s discourse during computer
game playing, Johansson (2000) noticed that the pronoun jag (‘I’)
can switch between two different meanings; either it refers to the
character that is controlled in the computer game, or it refers to the
player as an individual. Linderoth (2004) also studied the discourse
of Swedish children while playing computer games. He based his
analysis on Goffman’s idea that activity always takes place in a
frame, a certain context which, among other things, gives the utterances a particular meaning. Linderoth identified three meanings of
jag (‘I’), depending on which frame it occurs in. First, the pronoun
could refer to the speaker as a person—in the same way as noted by
Johansson (2000). Linderoth’s second meaning of jag (‘I’) is similar
to Johansson’s second meaning: it may refer to the character in the
game as a game unit. Third, Linderoth found an additional meaning,
which is that jag (‘I’) may refer to the character as a role. The distinction between the second and third meanings can be seen with
the following examples: If a person playing the board game Monopoly says I ended up on Park place she is not pretending to be either a dog, a shoe, or any of the other tokens in Monopoly (Linderoth, 2004). But if a person playing a computer game says oh,
someone is shooting at me in a squeaky voice, that person may be
acting as if she was in fact the character in the game.
While Johansson (2000) seems to view the multiple meanings of
jag (‘I’) as tied exclusively to the activity of playing computer games,
Linderoth (2004) acknowledges that the multiple meanings extend
to other games in addition to computer games. Indeed, the phenomenon does not have anything to do with computers per se. For
instance, consider the following example from a report of a golf
tournament on television (heard by the author): Woods hamnade i
vattnet (‘Woods ended up in the water’). Of course, we do not
Perspective on actions and events 201
wonder whether Tiger Woods became wet, but understand that it
was his golf ball that ended up in the water. In this case, it would be
perfectly reasonable to hear Woods say I ended up in the water.
Here, I does not refer to himself as an individual, but to an object
under his control.
Neither Johansson (2000) nor Linderoth (2004) gives an explanation why it is that jag (‘I’) can have multiple meanings. Such an explanation is given by Wilhelmsson (2001) in his conceptual framework for cognitive aspects of computer game activity, where he uses
the work of Lakoff on the conceptual structure of the self. According to Lakoff (1996), the self is conceptualised as consisting of two
parts: the Self, which is the part carrying out actions and associated
with social roles and past actions, and the Subject, who is the locus
of subjective experience, reasoning, and feeling. Wilhelmsson characterises the Game Ego as a Self that extends into the computer
game: ‘The Game Ego is the agency that exerts force upon the game
environment and that the game environment exerts its force upon’
(Wilhelmsson, 2001, p. 150). However, Wilhelmsson supplies no
empirical evidence of the existence or structure of the Game Ego.
Evidence of this kind can be provided by the present study of perspective. Wilhelmsson emphasises the role of motor activity in establishing a Game Ego: ‘If the game player is given motor control of
a character’s or an object’s motion, he or she is more likely to identify with that character or object’ (Wilhelmsson, 2001, p. 155) and
‘Controllability is . . . essential to the theory proposed. . . . Motor
activity and cognition add up to a Point of Being—a way of perceiving oneself as Being within the game environment’ (Wilhelmsson,
2001, p. 160). In the present study, a computer game involving little
or no motor activity, Anchorhead, was used and the resulting perspectives of the participants were studied, thus providing a test of
Wilhelmsson’s claimed importance of motor activity. Wilhelmsson
also discusses computer games without direct control, like the one
used in the present study: ‘This type of interface delays the action
and generates a weak motor link between the player and the Game
Ego’ (Wilhelmsson, 2001, p. 162). By hypothesis, this would then
result in a weak Point of Being, that is, not experiencing oneself as
202 Participating in a story
‘in the game’. Whether indirect control results in a weak feeling of
being ‘in the game’ is also answered within the present study of perspective.
Wilhelmsson (2001) also makes another reflection concerning
perspective in textual interactive fiction:
[The computer game] Zork [similar to Anchorhead] is sometimes referred to as a Second person game. This is so because of the use of
‘you’. . . . It is not a ‘you’ that performs the action. . . . When you
think about the game and the game environment you will refer it to
your Self and you[sic] Subject (Wilhelmsson, 2001, p. 173)
This conclusion is not based on systematic empirical studies, but
rather on intuition. Does it hold up to empirical testing?
A method to reveal perspective is to study language when people
talk about events. Both the verbal and the non-verbal parts of language can give clues to which perspective is adopted by a person
speaking about an event. As mentioned, verbal clues are, for example, choice of referent terms. Non-verbal clues can be found in gesture, particularly the viewpoint expressed in iconic gestures
(McNeill, 1992). For example, showing gesturally an action as if you
are performing it yourself, compared to gesturally illustrating the
action seen from an observer viewpoint.
When studying what perspective is used by people when talking
about events in participatory stories, the interviews (described in
Chapter 5) were used as data. In the data, there are four additional
conditions for comparison where people talk about events from
other sources: personal experience, laboratory tasks, a short story,
and a special non-participatory version of the participatory story.
The last condition provides a case for direct comparison against the
participatory story, since the events are the same, and described in
the same way, but without the element of participation.
The former research question can then be rephrased in terms of
language: What perspective is used when speaking about events
from participatory stories, and how does this differ when talking
about events from other sources, such as non-participatory stories
Perspective on actions and events 203
and real-life events? This question separates in two main parts, a
non-verbal (gesture) part and a verbal part. Two individual studies
will now be presented which study these approaches to perspective,
starting with how perspective is manifested in gesture.
8.1
PERSPECTIVE AS MANIFESTED IN GESTURE
When people retell events, gestural activity with the hands and arms
often spontaneously accompanies their speech. Several gesture
classifications systems have been developed, for instance, by Ekman
and Friesen (1969) and McNeill (1992). McNeill has carried out extensive studies of retelling of narrative and shown that gesture can
give important information about what speakers are thinking, especially if this information is only supplied through gesture and not in
speech. The gesture categories of McNeill’s classification are metaphorics, deictics, beats, emblems, and iconics. Metaphoric gestures give
something non-physical a physical shape. For instance, time is
shown as a physically extended object with points located to the left
(before) and right (after) of a centre (now). Deictic gestures are pointing gestures. They could point out a direction or a real or imaginary
object or person. Beat gestures are short, marking movements where
the hands quickly return to the rest position. Beats can be overlaid
onto iconics and metaphorics. Beats are motorically simple, but
cognitively complex. Occurrence of a beat sends the listeners in
search of another context. They function to introduce new characters and mark functions or attributes of an object talked about. Emblems are conventionalised gestures, such as making a V sign for
victory or giving ‘thumbs up’ to signal that something is good. Emblems are usually made with conscious effort and are different from
the other gestures in being more verbal, sometimes replacing words
or phrases. Thus, they do not, in the same manner as the other gestures, provide a window to the ongoing cognitive processes. Finally,
iconic gestures show a similarity to the things they depict, or they
show the manner in which something is performed.
204 Participating in a story
Part of McNeill’s gesture theory concerns perspective. For
McNeill, the perspective taken on a described event means the distance with which the speaker positions herself in relation to the
event. Gesturally, perspective is manifested in iconic gestures.
McNeill identifies two possible perspectives in iconic gestures: character viewpoint and observer viewpoint, and in support he mentions
confirming studies of retelling of narrative carried out by Stephens
and Tuite. Character viewpoint shows closeness to the events described:
Consider the event in the cartoon where Sylvester climbs up the pipe.
This could be conveyed gesturally in either of two ways. One would
be to move one’s arms up and down, as if climbing a ladder. Here, the
viewpoint would be the character’s: we imagine ourselves playing the
part of Sylvester—the pipe is in front of us and we move our hands up
and down as if clambering. . . . With this viewpoint we feel that the
narrator is inside the story. (McNeill, 1992, pp. 118–119)
In contrast, observer viewpoint expresses a distance to the event described:
A different gesture for the same event would be to make the hand into
Sylvester as a whole and cause it to rise upward. We see Sylvester before us, rising, upward, but we are not part of the scene. . . . With this
viewpoint, the narrator keeps some distance from the story. (McNeill,
1992, p. 119)
In the present study, the term ‘character viewpoint’ was replaced
with ‘subjective viewpoint’. The reason is that the data from the
present study contain people speaking about events that are sometimes fictional and sometimes real. McNeill’s term ‘character’ is
suitable for his studies of retelling of cartoon narratives, which are
fictional. But when people are talking of what they themselves did
at an earlier time, ‘character’ is not an appropriate term. Instead, the
general term ‘subjective’ was chosen, which reflects the fact that the
viewpoint is neutral with respect to fictional status. The speaker
could either be herself or himself, or could pretend to be someone
Perspective on actions and events 205
else acting out from that character’s perspective. This discussion is
omitted entirely by McNeill (1992). However, this is mostly a
change in terminology, since the categories character viewpoint and
subjective viewpoint in practice appear to cover the same gesture
occurrences.
8.1.1 Research question
If the audience in participatory stories ‘identifies’ with the agent in
the participatory story—if the player character is a ‘Game Ego’ by
means of a ‘projected Self’, in Wilhelmsson’s (2001) words (see
Chapter 4)—the audience would then experience, encode, and remember events from a first-person perspective. Thus, the prediction
was that gesture would show a higher proportion of subjective
rather than observer viewpoint in retellings of events from participatory stories, compared to the short story and the nonparticipatory version of the participatory story.
8.1.2 Method
8.1.2.1 Data
Gestural perspective on actions and events was investigated by using
the transcriptions of the interviews with the eight participants as
data. For description of participants, material, procedure, transcription, and coding, see Chapter 5. The interviews concerned talk
about events from five source conditions:
•
Game: Events from the computer game Anchorhead which
the participants played.
•
Story: Events from a short story which the participants
read.
•
Personal: Events that took place earlier in the participant’s
life.
•
Tasks: Events from three practical tasks carried out in a
laboratory setting.
206 Participating in a story
•
Non-participatory: Events from a special, adapted nonparticipatory version of the computer game, printed on
sheets of paper, which the participants read.
8.1.2.2 Transcription
The participants’ gestures were transcribed using a system based on
McNeill (1992). McNeill’s system suits the present study particularly well for two reasons: It was developed within the context of
narrative discourse, and it is part of a theory of the connection between cognition and gesture. Transcription proceeded as follows.
First, the arm and hand movement was classified as gesture or nongesture. A non-gesture would be, for example, scratching one’s chin.
If it was indeed a gesture, the following features were transcribed:
•
left or right hand
•
fingers: extended, fist, pointing
•
palm up or down
•
movement of arm and hand: start and end, trajectory
The detailed temporal unfolding of single gestures was not transcribed. McNeill (1992) identifies preparation, stroke, and retraction
phases of the gesture. This was considered too high a level of detail
with regard to the research questions asked in the present study. In
the transcriptions, it is mainly what corresponds to McNeill’s
stroke that is marked. For the transcription, it was not considered
important exactly where gestures start and end in relation to speech.
The timing of the gesture was marked in accordance with the Conversation Analysis standard, that is, placing it within double parentheses and aligning it with the speech that is concurrent with it.
These modifications follow the recommendations made by McNeill
(1992) regarding a simplified transcription, adapted to the purposes
of the present study.
The interviewer was not filmed, and consequently no gesture
transcription was made for the interviewer.
Perspective on actions and events 207
8.1.2.3 Coding
The gesture system of McNeill (1992) was used to classify the gestures in the data into the categories iconics, metaphorics, deictics,
beats, and emblems. The viewpoints of iconic gestures were classified as observer or subjective (or N/A if a viewpoint was missing or
could not be decided). Table 8.1 gives an overview of the coding
scheme for gestures. Figure 8.1 shows an example from the data of a
subjective viewpoint in an iconic gesture. An example of an observer viewpoint from the data can be seen in Figure 8.2.
An important clarification is made by McNeill (1992) concerning
the gesture coding system he proposes when he states that gesture
categories are not mutually exclusive. A gesture can belong to more
than one category. An iconic gesture showing how something
moves can also be deictic, showing the direction of movement.
Beats are often superimposed on other gestures. However, in the
present study, effort was made to find the most likely category to
which each gesture to be coded belonged. Every gesture was coded
with only one category.
Lastly, it should be noted that gesture coding (similar to much
other coding work) has a large arbitrary element. Coding demands
much experience, and even then, coding can differ substantially between coders (McNeill, 1992). In order to test coding reliability,
two transcriptions of the interviews with the eight participants were
randomly selected; one from participants 1–4 and one from participants 5–8. The transcriptions were used in full—and not partially—
in order to test coding for all the five conditions. The transcriptions
were additionally coded by a gesture researcher, unaware of the present research questions. As measures of inter-rater reliability, percent agreement and Cohen’s kappa were calculated for both gesture
categories and iconic gesture viewpoints.
208 Participating in a story
sen [så klängde (jag) mot (.) mot eh mot väggen]
then [(I) clung onto (.) onto eh onto the wall]
[((arms extended forward up fingers spread slightly
bent palms forward))]
Figure 8.1. Example of subjective viewpoint in an iconic gesture. See Chapter 5 for
transcription key for speech.
Perspective on actions and events 209
å
[från innergården tog jag mig uppåt]
and [from the backyard I went up]
[((right index finger in trajectory from right to left, then
upwards))]
Figure 8.2. Example of observer viewpoint in an iconic gesture.
210 Participating in a story
Table 8.1. Coding scheme for gesture.
Gesture
type
Iconic:
Observer
Description
Example
The gesture depicts how something appears
or in what manner an action is performed.
The gesture expresses an outside, observer
perspective—a narrator who ‘looks down’
on the events and describes from the outside.
‘I walked up the mountain’ while the person’s
finger traces a twisting
line in the air.
Iconic:
Subjective
The gesture depicts how something appears
or in what manner an action is performed.
The gesture expresses an inside perspective.
The person gesturing is taking the perspective of an agent who performs the action.
‘I walked up the mountain’ while the person
moves the arms as if
walking.
Iconic:
N/A
The gesture depicts how something appears
or in what manner an action is performed.
No viewpoint was present, or no viewpoint could be determined.
Deictic
The gesture points to a direction or a real
or imaginary object or person.
The index finger points
left while the person says
‘over there’
Metaphoric
Something non-physical is given physical
shape, e.g., length, front/back, up/down.
‘Forward in time’ is
shown by pointing to a
position to the left and
‘backward in time’ is
shown by pointing to a
position to the right.
Beat
Short, marking movement, where the
hand(s) quickly resume to the resting position. Sometimes occurs several in sequence.
Can be layered on top of iconics and metaphorics—then the hand(s) return to the
local resting position (which then is the
position of the gesture on which the beat is
layered).
‘It didn’t [matter]’
The hands move briefly
up, then are dropped
back into the lap.
Emblem
A conventionalised gesture.
Makes a V sign for victory.
Note: Adapted from McNeill (1992).
Perspective on actions and events 211
8.1.2.4 Analyses
For analysis of gesture, only cases of reference to past events were
included. In other words, the analysis was made exclusively on ‘the
level of narrated events’ (McNeill, 1992). This included references to
the game world (for the participatory story condition and the nonparticipatory version condition), the story world (for the short
story condition), and the situation of the original event (for the personal experience condition and the laboratory tasks condition).
Among the other levels present in the data, was, for instance, the
conversational level, which had utterances such as ‘I remember now
that…’. On the conversational level, the speaker uses I to refer to
herself and the verb is in the present tense. Since the conversational
level was present for all conditions, it does not reveal how the perspective on actions and events in the conditions may be different.
Talk on the conversational level together with talk about other
things besides events from the five sources was disregarded in the
analyses.
During coding of gesture, it became evident that also gestures
other than iconic gestures seemed to express a viewpoint. This led
to an alternative gesture viewpoint coding, where all gestures were
coded according to the scheme above. Coding viewpoint in gestures
other than iconic gesture is not taken up by McNeill (1992) in his
gesture theory, and it therefore has less theoretical underpinning.
However, there are at least two arguments in support of carrying
out this alternative analysis in the present study. First, excluding
these other gestures from the analysis of viewpoint would not reveal
the complete picture of what viewpoints were being communicated
through gesture. Second, gesture coding proves to be quite arbitrary. An iconic gesture could just as easily be coded as a deictic or
metaphoric gesture, thereby excluding it from further analysis of
perspective. By coding viewpoint for all gestures, the chances are
higher to capture any viewpoints that were expressed gesturally.
For each participant, the total length for each of the five conditions was calculated by summing up that condition’s segments (see
Table 5.7). The gesture occurrences were divided by the total length
to give occurrences per minute. Finally, to reduce error variance in
212 Participating in a story
the analysis, for each participant, a ratio of subjective and observer
viewpoint was used instead of the actual frequencies. The average of
all participants was used in the analyses.
8.1.3 Results
Results are presented first for the general occurrences of gesture,
showing all types, then for the iconic gesture viewpoints.
8.1.3.1 General occurrences of gesture
The results of analysis of inter-rater reliability for the gesture coding
showed that percent agreement was 56.1 percent and Cohen’s kappa
was .318.
The occurrences per minute of gesture types (average for all participants) for the five conditions are shown in Table 8.2. There, it can
be seen that iconic gestures constitute between 24.6 and 47.0 percent
of the total gestures, depending on condition (40 percent averaged
across all conditions).
Table 8.2. Occurrences of gesture types per minute in the five source conditions, for
narrative level.
Source condition
Gesture type
Game
Story
Personal
Tasks
Nonpartic.
Deictics
1.05
0.23
0.45
0.55
0.74
Iconics: subjective
0.93
0.47
0.58
5.70
0.47
Iconics: observer
1.45
0.73
1.59
0.20
1.37
Iconics: N/A
0.17
0.13
0.06
0.00
0.10
Metaphorics
1.02
0.81
0.48
2.84
2.13
Beats
1.36
0.43
1.99
3.34
2.71
Emblems
0.03
0.03
0.24
0.68
0.38
Total gesture freq.
6.01
2.83
5.39
13.31
7.9
Percent iconics
42.4%
47.0%
41.4%
44.3%
24.6%
Perspective on actions and events 213
8.1.3.2 Gesture viewpoints
The test of inter-rater reliability for the coding of viewpoints
showed that percent agreement was 78.3 percent and Cohen’s kappa
was .566.
The analysis of viewpoint in iconic gesture and the alternative
coding procedure (where viewpoint was coded for all gestures)
yielded similar results. The proportions of observer and subjective
viewpoints of iconic gestures can be seen in Figure 8.3. An analysis of
variance of the viewpoint proportions across the five conditions
showed that there was a difference in means, F (4,15) = 8.416,
p < .002. Bonferroni post hoc test with the alpha level at .05
showed that the Tasks condition had a higher proportion of subjective viewpoints than the other four conditions. The proportions of
observer and subjective viewpoints did not differ in any of the other
four conditions. The increase in subjective viewpoint for the Task
condition was dramatic, lowering the proportion of observer viewpoint to a few percent. The alternative viewpoint coding, in which
viewpoint was coded for all gestures (not just iconics) gave similar
results, so these data are not presented here. The gestures that were
judged to express viewpoint were, in addition to iconics, metaphorics, deictics, and emblems. Here too, as in the analysis of iconic
gesture only, an analysis of variance showed a difference in means of
viewpoint proportions across the five conditions, F (4,15) = 6.151,
p < .01, and Bonferroni post hoc tests with alpha level at .05 revealed that the Tasks condition had a higher proportion of subjective/observer viewpoints (no other differences in means)—the same
dramatic high proportion of subjective viewpoints as yielded by the
analysis of only iconic gestures.
214 Participating in a story
Proportions of subjective and observer viewpoints (iconic gestures)
100%
50%
0%
Game
Story
Personal
Tasks
Non-part.
Source
Observer
Subjective
Figure 8.3. Proportions of subjective and observer viewpoints in iconic gestures on
the level of narrated events. (Non-existent viewpoints or cases where viewpoint
could not be decided are not shown.)
8.1.4 Discussion
The analysis of perspective as manifested in gesture in terms of proportion of subjective and observer viewpoints showed no differences between the Game, Story, Personal, and Non-participatory
conditions, but in contrast to these four conditions, the Tasks condition had a higher proportion of subjective viewpoints. For the
four conditions Game, Story, Personal, and Non-participatory, observer and subjective viewpoints occur about equally often. This
suggests that the events are thought of both from the ‘outside’ and
from the ‘inside’, that is, the participant sometimes conceptualised
Perspective on actions and events 215
as being outside the event, and sometimes as taking part inside the
event.
8.1.4.1 Inter-rater reliability
The coding of gesture turned out to be somewhat unreliable—which
is also an experience that has been reported from earlier studies
(McNeill, 1992). It is difficult to operationalise the gesture categories
and variation in experience with gesture coding can result in variable coding among researchers. However, most importantly, the
coding was satisfactory for viewpoints in iconic gesture—and these
data were the basis for analysis of gesture perspective.
8.1.4.2 Comparison with earlier research
Comparing the proportions of gesture types when describing events
on the narrative level (Table 8.2) with figures presented in McNeill
(1992) shows that the proportions are roughly similar. McNeill reports 57 percent iconics, 34 percent beats, 6 percent deictics, and 3
percent metaphorics. A few differences from the present study are
noteworthy. First, the present study obtained a higher number of
metaphoric gestures throughout the five conditions. Second, the
Game condition yielded twice as many deictic gestures as any of the
other conditions. This may be an indication of the spatial nature of
the computer game. However, it should be noted that the conditions differ between the present study and McNeill’s study. Also,
slight differences in coding between the present study and McNeill’s
study may influence the proportions.
Turning next to the viewpoints, McNeill (1992) reports from
studies by Church and colleagues on retelling of cartoon narratives,
where the iconic gestures in 40 percent of the cases express an observer viewpoint and 60 percent of the cases express a character
viewpoint (translated to subjective viewpoint in the present study).
This relates perhaps most closely to the Story and Nonparticipatory conditions in the present study. Even though the two
viewpoints occur equally often in these two conditions in the pre-
216 Participating in a story
sent study, the figures are close to those obtained by Church and
colleagues.
8.1.4.3 Why not more subjective viewpoints in personal experience?
The equal proportions of viewpoint is perhaps most surprising in
the case of personal experience, since one would expect that events
which actually involved the participants would be viewed to a
higher degree from the ‘inside’—expressed using subjective viewpoints (McNeill, 1992).
A factor that may explain why there was a difference between
the Personal and Tasks conditions—two conditions which would be
expected to show similar results because both are personally experienced, real events—is that different amounts of time passed since the
events took place. In research on visual perspective in episodic
memory, it has been found that recent events tend to be remembered using a field perspective (as if seeing out of one’s own eyes)
while distant events tend to be remembered using an observer perspective (Nigro & Neisser, 1983; Robinson & Swanson, 1993;
McIsaac & Eich, 2002). If remembered perspective is assumed to be
manifested also in gesture, the amount of time passed could be part
of the explanation, but it does not seem to fully account for the results. The large difference found in the present study—subjective
viewpoint outnumbering observer viewpoint by a factor of 30—has
not been found in studies of remembered perspective (Nigro &
Neisser, 1983; Robinson & Swanson, 1993; McIsaac & Eich, 2002).
8.1.4.4 Game versus Non-participatory
Also when comparing the Game condition with the Nonparticipatory condition, the results are surprising. Since the Nonparticipatory condition to a lesser extent involves participation, it
would be expected that the events in retelling show fewer occurrences of subjective viewpoints, but this was not what was found.
Let us consider some alternative explanations of the results.
Perspective on actions and events 217
8.1.4.5 Alternative explanations
A possible explanation of the results would be that subjective viewpoint is correlated with feeling of involvement, and not with
whether the person actually participated or not, in the events talked
about. This would explain the results for the Game condition,
which one is somewhat involved in. One is less involved in the
short story, and even less involved in the non-participatory version
of the computer game (this is consistent with reports from participants during the interview that they thought the non-participatory
story was a bit odd). The laboratory tasks would be highly involving, since they are carried out ‘directly’ in the ‘real world’ by the
participants. However, the Personal condition seems to present an
anomaly under this explanation, which would be expected to have a
relatively higher degree of subjective viewpoint gestures, but the
results show none of this. Are not people involved in things that
happened to them personally some time ago? Again, the finding
concerning memory age from research on phenomenal perspective
in remembered events may help explain this. As the events become
more distant in time, the portion of observer perspective increases
(Nigro & Neisser, 1983; Robinson & Swanson, 1993; McIsaac &
Eich, 2002).
A related possible explanation is that use of subjective viewpoint
is correlated with degree of problem-solving carried out by the participant. The short story, the non-interactive version of the computer game, and personal experience involves a relatively low degree
of problem-solving activity. The situation with the laboratory tasks
was framed in a problem-solving context and the high degree of
subjective viewpoint gestures mirrors this. If this explanation is accepted, one is forced to view playing the computer game as low in
problem-solving activity.
The extreme result concerning the Tasks condition shows that
the events talked about from the laboratory tasks are viewed almost
exclusively from a subjective viewpoint. How can this result be explained? A difference in viewpoint because of varying communicative demands in the five conditions can be ruled out—the situation
was virtually identical for all conditions (see Chapter 5). Another
218 Participating in a story
explanation of the observed frequencies of observer and subjective
viewpoints is that they are correlated neither with participation nor
with involvement, but result from the nature of what needs to be
communicated. The description of certain actions may demand an
observer viewpoint while other actions may demand a subjective
viewpoint. This may have nothing to do with who carried out the
action: the participant or someone else. Particularly, taking an observer viewpoint may be necessary in order to gesturally show actions which involve space that lies beyond the reach of the speaker’s
arms. That way, a real space is compressed into a smaller space,
which serves as a model to illustrate the actions which took place.
This leads the speaker to adopt an observer viewpoint. Actions may
be described using subjective viewpoint gestures as far as possible,
and only if events involve space larger than that around the body,
gestures with observer viewpoint are used. Reviewing the results
from the five conditions shows that they are consistent with this
explanation. The laboratory task condition encompassed events
which could be well described gesturally using a subjective viewpoint, since they involved only a limited space around the body.
The short story, the non-participatory version of the computer
game, and personal experience all in addition involved events that
took place in a larger space, which forced the participants to use an
observer viewpoint as well. This explanation requires us to see the
computer game events not as something occurring in front of a
computer on a desk, but as events in the game world.
The original hypothesis that events that one participated in lead
to a higher degree of subjective viewpoint can be grouped together
with the hypotheses that feeling of involvement or problem-solving
activity increases subjective viewpoint. This group of explanations
could be labelled the ‘participation’ explanation. The other explanation, that certain actions—because of their spatial properties—
require observer viewpoint whereas other actions require subjective
viewpoint, regardless of participation, can be called the ‘nature of
actions’ explanation. There seems to be no way of teasing apart the
participation and the nature of actions explanations in the present
Perspective on actions and events 219
study. Both have to be considered possible explanations of the results.
Next, an analysis of perspective as manifested in speech is presented which clarifies the picture of the perspective that participants
adopt when talking about actions and events from the participatory
story.
8.2
PERSPECTIVE AS MANIFESTED IN SPEECH
When talking about a past event, a speaker has a lexical choice of
how to refer to the animate agents involved in the event. For example, ‘I came’, ‘he came’, and ‘one came’ may describe the same
event. This choice of terms for the referents is one type of choice of
perspective, and will here be called agent perspective. Agent perspective reveals the speaker’s view of the events and gives clues to ongoing cognitive activity. Let us exemplify this by looking at the agent
perspectives possible in Swedish when one is talking about past
events.26 The first-person singular pronoun jag (‘I’) may be used, for
instance jag tog en promenad i morse (‘I went for a walk this morning’). This choice of pronoun indicates a closeness to and signals
participation in the event talked about on the part of the speaker.
The second-person singular pronoun du (‘you’) signals lack of
participation and distance on the part of the speaker, as in du tog en
promenad i morse (‘you went for a walk this morning’). (It is possible that the speaker also took part in this event, but it is not signalled by the choice of pronoun.) The pronoun du may also be used
(loosely) to refer to the speaker, as in det var så mörkt att du inte
kunde se handen framför dig (‘it was so dark you couldn’t see your
hand in front of you’). In this case, the speaker uses the secondperson singular to refer to an unspecified person being in the same
situation as the speaker was. In this usage, du means almost the same
Since the data in the present study consist of people talking about past events, the
discussion in this chapter is delimited to ways of referencing agents of past events.
26
220 Participating in a story
thing as jag, but it still signals distance and one cannot infer that the
speaker participated in the event.
The use of a third-person singular pronoun signals distance of the
speaker from the event and does not show that the speaker participated in the event—for example, han tog en promenad i morse (‘he
went for a walk this morning’). The same distant perspective is expressed by using nouns, such as mannen (‘the man’) or Michael.
Another way of referring to the agent of some event is by using
the indefinite pronoun man, as in man kunde inte tro sina ögon (‘one
couldn’t believe one’s eyes’). There is no single English equivalent
to this Swedish pronoun (Andersson, 1972; Norell, 1995)—man can
be translated into ‘one’, ‘you’, or ‘they’, or expressed using other
constructions such as passive, depending on the meaning. Andersson (1972) claims that man has two usages in Swedish: one general
use, where reference is to the domain of all people, and one anaphoric use, where the reference is to a subset of the domain of all
people. Andersson gives as an example of the first case man skall inte
tro allt som sägs på TV (‘one/you should not believe everything that
is said on TV’). Here, the speaker is included in the domain of people referred to. Andersson’s example of the second case is på TV har
man fått en ny chef (‘on TV they have got a new boss’). Here, the
speaker is not included in the domain of persons referred to. Regardless of usage, man can be said to distance the speaker from what
is being said.
Describing an event using the pronoun vi (‘we’)—which can be
viewed as the plural version of jag (‘I’)—signals closeness to the
events. It does however express a somewhat weaker participation in
the events on the part of the speaker compared to jag. Sometimes vi
is used as in vi vann världscupen (‘we won the World Cup’), if vi
refers to the nation to which the speaker belongs or to the group of
people with the same nationality as the speaker, but the speaker
needs not be part of the group that is referenced in a literal interpretation.
The pronoun dom (‘they’) (including the forms de and dem) is the
plural form to refer to the third person. It signals neither closeness
to nor participation in the events talked about.
Perspective on actions and events 221
Table 8.3 summarises the agent perspectives possible in Swedish
when talking about past events, and what they signal in terms of
proximity to and participation in the events.
Table 8.3. Possible agent perspective when talking about past events, and what they
signal in terms of proximity and participation.
Grammatical
categories
Term for
referent,
Swedish
Term for
referent,
English
Proximity of
speaker to
events
Signals
speaker
participation
First-person singular
pronoun
jag
I
closeness
yes
Second-person
singular pronoun
du
you
distance
no
Third-person singular
pronouns and nouns
han, hon
…
he, she,
nouns
distance
no
First-person plural
vi
we
closeness
mostly
Third-person plural
de/dom
they
distance
no
Indefinite pronoun
man
one/you/
they/…
distance
sometimes
8.2.1 Research questions
Which agent perspectives occur when participants talk about events
from the five conditions? More specifically, how do people refer to
the player character in the computer game—as ‘I’ or ‘she’ or in some
other way? Based on Wilhelmsson’s (2001) analysis of Game Ego, it
was predicted that the participants referred to the player character
in a way similar to personal experience, that is, using the pronoun
jag (‘I’).
8.2.2 Method
A qualitative analysis was carried out of which agent perspectives,
out of the possible ones discussed above, occur when people talk
222 Participating in a story
about events from the five sources in the study, followed by a quantitative analysis of agent perspective frequencies. This lets us see
how people talk about what happens in the participatory story, and
how people referred to the agent (player character) in the participatory story.
Parts of the method were similar to the method used to analyse
gesture and a presentation can be found in that section above.
In analysing perspective as expressed in speech, data from the interviewer as well as uncertain speech (transcribed within parentheses) were disregarded.
8.2.2.1 Coding of agent perspective
The participants’ utterances were searched for terms fulfilling three
criteria; the term should
•
refer to an animate agent (the speaker, a character in a story
or a game, etc.)
•
refer to someone who performs an action
•
occur on the level of narrated events (in the case of the
computer game: references to the game world and its objects, etc., in the case of the short story: references to the
story world, and so on).
In the group of third-person singular pronouns and nouns, pronouns such as ingen (‘nobody’) and others which expressed a common perspective with that class were included. The pronoun du
(‘you’) and the construction duet (‘the you’) had only a single occurrence each in the data and were therefore discarded from further
analysis. When instances of repeated occurrences of perspective expressions were found in the data (such as stuttering-like parts of
utterances), only one of the occurrences was counted. Instances of
agent perspectives that occurred in interrupted speech, where it
could not be decided what the participant intended to say, were
disregarded.
Perspective on actions and events 223
8.2.2.2 Pronoun switches
As it was noted that participants sometimes switched between jag
(‘I’) and man (‘one’/‘you’) in talk about the computer game, these
switches were tracked. A switch was defined as having taken place
when there was reference to the speaker (in the playing situation) or
to the player character with jag in one section of the discourse, and
then in another section of connected discourse, shift to the use of
man (or vice versa), where connected discourse is discourse with a
unified theme, not interrupted by the interlocutor except for feedback utterances or signals. Switches were counted and the (probable) reason for the switch was noted, as determined from the surrounding discourse.
8.2.3 Results
Results for occurrences and proportions of agent perspective are
presented first, followed by statistical comparisons among proportions. Then, a presentation follows of how the agent perspectives
were used when talking about events from the five sources. Finally,
results of analyses of switches between the first-person pronoun jag
(‘I’) and man (‘one’/‘you’) are presented.
8.2.3.1 Occurrences and proportions of agent perspective
The occurrences and percentages of agent perspectives (occurring on
the level of narrated events) expressed by the participants can be
found in Table 8.4 (percentages add up to one hundred for each
condition). The proportions are presented graphically in Figure 8.4.
From Table 8.4, it can be seen that talk about events from the
participatory story, the short story, and personal experience contained all agent perspectives presented in the table. Talk about the
non-participatory version included all agent perspectives except
first-person plural (vi/‘we’). Talk about events from the laboratory
tasks contained only the agent perspectives first-person singular
(jag/‘I’) and the indefinite pronoun man (‘one’/‘you’).
224 Participating in a story
Proportions of verbal agent perspectives
Dom
Dom
Vi
Dom
Vi
Man
Dom
Man
Man
Man
Man
Han/hon/nouns
Han/hon/nouns
Han/hon/nouns
Han/hon/nouns
Jag
Jag
Jag
Game
Tasks
Personal
Jag
Jag
Story
Non-part.
Condition
Figure 8.4. Proportions of agent perspective in talk about events from the five
sources (average for all participants).
Table 8.4. Agent perspective in talk about events from the five sources (average for
all participants): number of occurrences and percentages.
Agent perspective
1st p. pron.
jag
3rd p. sing.
e.g., hon
Indef. pron.
man
Source
#
%
#
%
#
Game
192
37.9
17
4.7
149 55.4
%
1st p. pl.
vi
3rd p.
pl. dom
#
%
#
%
1
0.4
8
1.6
Tasks
60
64.1
0
0
29
35.9
0
0
0
0
Personal
76
48.5
9
10.1
31
13.9
17 26.6
3
0.8
Story
6
2.6
126 59.7
7
3.5
1
0.3
71 33.9
Non-part.
2
1.3
91
6
5.2
0
0
33 25.2
68.3
Perspective on actions and events 225
Comparing overall patterns of agent perspective, it can be seen from
Figure 8.4 that the results form two main groups. The Game condition is most similar to the Personal and Tasks conditions. The second group is constituted by the Story and Non-participatory conditions, which show similar patterns of agent perspective. Interestingly, talk about events from the participatory story and the nonparticipatory version of the same (these two had identical textual
descriptions) showed completely different patterns of agent perspective, the former dominated by jag (‘I’) and man (‘one’/‘you’), while
the latter dominated by the agent perspective group consisting of
singular and plural third-person pronouns, and nouns.
8.2.3.2 Comparing proportions of agent perspectives
Game condition. Within the Game condition, there was a difference in means of proportions of the five agent perspectives,
F (4,15) = 6.024, p = .004, and Bonferroni post-hoc tests with alpha
= .05 showed that the mean for the indefinite pronoun man
(‘one’/‘you’) was higher than the third-person singular group (p =
.032), vi (‘we’) ( p = .018), and dom (‘they’) ( p = .021). No difference was found between the proportions of man and jag (‘I’).
Tasks condition. Two agent perspectives were present in the
Tasks condition: first-person singular jag (‘I’) and the indefinite pronoun man (‘one’/‘you’), and there was no difference between these
proportions (Bonferroni post-hoc tests, alpha = .05).
First-person singular. Comparing the proportion of the firstperson singular perspective (jag/‘I’) across the five source conditions
revealed a difference in means, F (4,15) = 5.395, p = .007. Talk about
events from the laboratory tasks contained a higher proportion of
first-person singular agent perspective than talk of events from the
short story as well as from the non-participatory version (Bonferroni post-hoc tests, alpha = .05). Although means of the Game
and Personal conditions appear much higher than the Story and
Tasks conditions, Bonferroni post-hoc tests revealed no difference
(probably because of large variation between participants).
Third-person singular. The agent perspective that included thirdperson singular pronouns, and nouns, differed across source condi-
226 Participating in a story
tions, F (4,15) = 64.817, p < .001. Talk about events from the short
story and the non-participatory version had a higher proportion of
this perspective than did the other three source conditions (Bonferroni post-hoc tests, alpha = .05).
First-person plural. There was a difference among mean proportions of the first-person plural agent perspective (vi/‘we’), F (4,15) =
5.778, p = .005. Talk about events from personal experience contained a higher proportion of this perspective than either of the
other four source conditions (Bonferroni post-hoc tests, alpha =
.05).
Third-person plural. The proportion of third-person plural agent
perspective (dom/‘they’) differed across conditions, F (4,15) =
22.543, p < .001. Talk about events from the short story as well as
from the non-participatory version contained a higher proportion
of this perspective than the other three source conditions (Bonferroni post-hoc tests, alpha = .05).
Indefinite pronoun man (‘one’/‘you’). There was a difference in
mean proportions of the indefinite pronoun man (‘one’/‘you’)
agent perspective across conditions, F (4,15) = 3.430, p = .035, but
Bonferroni post-hoc tests (alpha = .05) did not successfully identify
the conditions in which these differences occurred. A visual inspection of the data makes it likely that the difference consisted of the
Game condition having a higher proportion of this perspective than
the Story and the Non-participatory conditions.
In summary, the comparisons of proportions across sources of
perspective as expressed in speech revealed two groups of patterns,
consisting of, on the one hand, the Game, Personal, and Tasks conditions, and on the other hand, the Story and Non-participatory
conditions. (The exception was that the proportion of the firstperson pronoun jag (‘I’) did not differ in the Game and Personal, as
compared to the Story and Non-participatory conditions.)
8.2.3.3 How was the agent perspectives used by the participants?
Let us now turn to how the agent perspectives were used by the
participants to talk about events from the five sources. Some occurrences of agent perspective concern the ‘main agent’ in the event
Perspective on actions and events 227
(such as the player character in the computer game, or the participants themselves in personal experience). However, there are also
other agents in the events described by the participants. For example, talk about personally experienced events involved other people
who also did things in the events, beside the participant. These
other instances also contribute to the total sum of occurrences of
agent perspective. Finding the referent of a verbal expression is an
act of interpretation. Background knowledge of the computer game
and the short story, as well as the interview as a whole, was used in
this process.
Since the agent perspective expressed by the indefinite pronoun
man (‘one’/‘you’) had an unexpectedly high occurrence in the
Game condition, examples of this agent perspective from the data of
the other conditions are presented for comparison throughout the
presentation of the use of agent perspective.
Computer game. In the Game condition, the third-person group
(han ‘he’, hon ‘she’, and nouns) and dom (‘they’) were used by one
participant to describe the background to the participatory story
(but was not used in talk about the actual events that took place).
The pronoun dom was also used to describe a group of non-player
characters in the game. The pronoun vi (‘we’) was used once to refer to the player character and her husband Michael. The latter was
also referred to by the pronoun han (‘he’). The pronouns jag (‘I’)
and man (‘one’/‘you’) were used to refer both to the player character and the participant herself/himself, although it was often
difficult to establish a unique referent. In Example 8.1, man seems
to refer to the player character, while it seems to refer to the participant himself in Example 8.2 (see Chapter 5 for a transcription
key).
(8.1)
Using man to refer to the player character (participant 1):
där man ställde en soptunna öppnade fönstret och kröp
where you put a trash can opened the window and crawled
in genom fönstret
in through the window
228 Participating in a story
(8.2)
Using man to refer to the participant (participant 1):
det var bara det att man tog tid på sig å sen så kollade man
you just had to take your time and then you checked things
upp och grejade å så
out and did things
Short story. In the Story condition, the third-person group (han ‘he’,
hon ‘she’, and nouns) and dom (‘they’) were used to refer to characters in the story. The indefinite pronoun man (‘one’/‘you’) was
used by two participants to describe how to carry out a signalling
procedure, which occurred in the short story, and by one participant to account for a general requirement holding in a situation in
the short story. These three occurrences did not have completely
unambiguous referents, but it was clear that they referred to the
group of characters in the short story.
Personal experience. In the Personal condition, jag (‘I’), vi (‘we’),
hon (‘she’) and nouns, and dom (‘they’) referred to the participant
and people around her or him. The indefinite pronoun man
(‘one’/‘you’) was used by one participant to add psychological distance, shown in example 8.3.
(8.3)
Use of man to add psychological distance (participant 3 and Pierre,
the interviewer):
Paul: att spionera [på] folk
spying
[on] people
Pierre:
[a] har du kan du berätta om nånting
[a] have you can you tell about something like
sånt (.) har du gjort nåt liknande eller har du
that (.) have you done anything like that or have you
vart med om
experienced
Paul: har jag säkert gjort när jag var liten å så där
I have surely when I was little
Pierre: mm
Paul: då smög man ju omkring och tjuvkika å så där på konstiga
then we did sneak around and peeked and stuff at the odd
gubben på gatan å sådär gick man å tjuvkika på
man on the street and stuff we went to peek at
Perspective on actions and events 229
The indefinite pronoun man was also used to be general (examples
8.4 and 8.5), to describe a procedure (example 8.6), to refer to an
indefinite group of people (example 8.7), and to communicate what
would be experienced by anyone in the situation described (example
8.8). Thus, use of man was not ambiguous.
(8.4)
Use of man to be general (participant 3):
jag var fascinerad av det som liten var jag att man
I was fascinated by that when I was little I was that you
kunde signalera morse (.) med en ficklampa till exempel
could signal in morse code (.) with a flashlight for example
då kunde man tala med en ficklampa
then you could speak with a flashlight
(8.5)
Use of man to be general (participant 4):
när man tappar om man tappar nyckeln eller man till
when you lose if you lose the key or when you for
exempel bara glömt nyckeln eller så
example just forgot the key
(8.6)
Use of man when describing a procedure (participant 4):
så här (.) att man har haft ett eh system där man haft en en
like this (.) that there was a system where there was a a
lapp som man drar med en tråd
note that you pull with a string
(8.7)
Use of man when referring to an indefinite group of runners (participant 4):
där man hade sprungit över som hade blivit helt
where they had been running which had been
söndertrampat så det var alldeles dyigt
tramped down so it was all muddy
(8.8)
Use of man when referring to anyone who would be in a situation
(participant 4):
försvann skon rätt i så man såg ett tomt hål mitt i dyn
the shoe disappeared so you saw an empty hole in the mud
230 Participating in a story
Laboratory tasks. In the Task condition, jag (‘I’) was used to refer to
the participant when describing what she or he did. The indefinite
pronoun man (‘one’/‘you’) was used to refer to what anyone in that
task situation would experience (example 8.9) or what one was supposed to do in the tasks (example 8.10). Also, there was one possible
use of man to add distance (example 8.11). As can be seen from the
examples, the reference of man was relatively clear.
(8.9)
Use of man to describe what anyone in the situation would experience (participant 6):
först så fick man (.) ett papper där man fick välja vilken typ
first you got (.) a paper where you could choose what type
av uppgift man ville göra om man ville göra (.) gissa eller
of task you wanted to do if you wanted to do (.) guess or
jämfö:ra eller mä:ta
compa:re or mea:sure
(8.10) Use of man to describe what one was supposed to do in the tasks
(participant 7):
första uppgiften var att man skulle (.) beräkna (.) ((smack))
the first task was to (.) calculate (.) ((smack))
öh arean eller a i rummet man satt i
eh the area or eh of the room you were sitting in
(8.11) Use of man to add distance (participant 7):
dom flesta hade man ju en klar bild av vad det var
for most of them you had a clear picture what they
i men det kan ju vara fel
contained but that could be wrong
Non-participatory version. In the Non-participatory condition hon
(‘she’), han (‘he’), alla (‘everyone’), ingen (‘no one’), and nouns, and
dom (‘they’) were used to refer to the woman and other characters
in the story. The indefinite pronoun man (‘one’/‘you’) was used to
refer to anyone who might be in a certain situation (examples 8.12
and 8.13).
Perspective on actions and events 231
(8.12) Use of man to describe what anyone in that situation would see
(participant 6):
det första man öh ser var att det var en sto:r eh svart dörr
the first thing you eh see was that it was a bi:g eh black
door
(8.13) Use of man to describe what anyone in that situation would see
(participant 8):
det såg stort ut utifrån men gav ett öh ett öh det gav ett
it looked big from the outside but have a eh a eh it gave an
tryckande intryck när man kom in att man kände sig eh (.)
oppressive impression once you came in that you felt eh (.)
nedtryckt
oppressed
And as also occurred in the Game condition, man (‘one’/‘you’) was
used to refer to anyone playing a game (example 8.14).
(8.14) Use of man to refer to anyone playing a game (participant 7, and
Pierre, the interviewer):
Marie: den var skriven ganska roligt för att det jag tänkte
it was written quite amusingly because what I thought
på en gång på (.) dataspel såna här
about right away was (.) computer games these
äventyrsspel
adventure games
Pierre: mm
Marie: å det var väldigt upplagt så att man skulle gå till nord
and it was very arranged so that one should go to north
nordväst å hon skulle gå till nordväst å så skulle hon så
northwest and she was going to the northwest and then
såg hon miljön å allting som var
she was then she saw the surroundings and everything
där då som man kunde använda sig av eller ta upp
there then that could be used or picked up
8.2.3.4 Switches between jag and man in the Game condition
A total of 46 switches between the first-person pronoun jag (‘I’) and
the indefinite pronoun man (‘one’/‘you’) were found in talk about
232 Participating in a story
events from the computer game: 22 shifts from jag to man and 24
shifts from man to jag. Example 8.15 shows a part of the interview
where there is a switch from man to jag.
(8.15) Use of man to refer to anyone playing a game (participant 7):
man får klättra in genom fönstret (.) sen får man
you have to climb in through the window (.) then you get
nycklarna där och sen så går man till (.) sen gick jag till
the keys there and then you go to (.) then I went to
puben (.) och där så hittade jag (en) flaska jag tog och
the pub (.) and there I found (a) bottle that I took and then
sen så gick jag till universitetet (.) och träffade min man
I went to the university (.) and met my husband michael
michael (.) och sen gick vi till huset (.)och gick och la oss
(.) and then we went to the house (.) and went to bed
Table 8.5 shows the reasons for the switch as determined by looking
at the functions of the pronoun immediately after the switch. The
total occurrences of jag and man were 341, which, given the 46
switches, makes on average a switch around every 7 occurrences of
these two pronouns.
Table 8.5. Switches between jag and man in the Game condition.
Switch
#
Function
jag → man
16
what would happen to any player
jag → man
3
what one was supposed to do
jag → man
1
thinking about a future possibility in the playing situation
jag → man
1
general reference to someone using a technique
jag → man
1
constraints of the game, holding for any player
specifics of what was done in the game, or could be done
man → jag
16
man → jag
4
thoughts about how to proceed in the playing situation
man → jag
3
mental state in the playing situation
man → jag
1
what happened in the playing situation, game as abstract object
Perspective on actions and events 233
8.2.4 Discussion
When analysing how perspective on actions and events from the
five sources was expressed in speech, the results revealed two main
groups. The first group of perspective use consisted of the Game,
Personal, and Tasks conditions. This group was characterised by the
use of jag (‘I’), vi (‘we’), and man (‘one’/‘you’). The other group
consisted of the Story and Non-participatory conditions. Here, the
use of perspective was characterised by the third-person groups of
singular (han ‘he’, hon ‘she’, and nouns) and plural (dom ‘they’). The
interpretation of these results is that participants mainly took an
outside perspective when talking about the events from the short
story and the non-participatory version. In the other three conditions, the perspective was mainly an inside perspective, but the
common use of the indefinite pronoun man (‘one’/‘you’), especially
in the Game condition, added distance to the perspective.
8.2.4.1 Agent perspective in the Game condition
The case of most theoretical concern in the results is when participants talk about events from the computer game. In contrast to the
other conditions, where the perspective is more or less given, the
events from the computer game present an intriguing case for studying how actions and events in the computer game are thought about
by the participants.
The indefinite pronoun man (‘one’/‘you’). Perhaps the most unexpected result concerning the use of perspective in the Game condition is the frequent occurrence of the indefinite pronoun man
(‘one’/‘you’) (55.4 percent, compared to 5.2 percent in the Nonparticipatory condition). What does man mean? In one sense, the
meaning of man seems to be rather close to jag (‘I’) in that these two
terms can sometimes be exchanged without changing much of the
meaning (se examples 8.1 and 8.2). Sometimes man was used in order to express something one should do, considering the intention
of the designer of the computer game. However, the event described
need not have occurred. Thus, using man was a way of referring to
actions which should be performed, even if they were not actually
234 Participating in a story
performed by the participant when playing the game. Used in this
way, man refers to a group of people of which the speaker is not a
part (cf. Andersson, 1972).
The high frequency of the indefinite pronoun man may be due to
the fact that Swedish was the language used to describe events from
the computer game. Perhaps other languages that do not have a direct corresponding indefinite pronoun (such as English) would have
led to a lower frequency of this agent perspective. This is a question
for future research.
There is sometimes referential ambiguity in the use of man, as in
example 8.16. Is ‘found’ something happening to the person or the
game character? This ambiguity is also present in the use of the firstperson singular pronoun, which will be discussed next.
(8.16) Referential ambiguity in the use of man (participant 1):
det enda man hittade överhuvudtaget var en: (.) et- en spade
the only thing you found at all was a: (.) a shovel
The first-person singular pronoun jag (‘I’). The first-person pronoun
jag (‘I’)—like the indefinite pronoun man (‘one’/‘you’)—was ambiguous and referred sometimes to the participant as an individual
and sometimes to the player character. It was not always possible to
decide to whom a particular instance of jag (‘I’) referred.
The two senses of jag (‘I’) noted by Johansson (2000)—the participant and the player character—were found also in the present study.
There are at least three differences between Johansson’s study and
the present study, however, namely, that Johansson studied children, that her discourse was recorded while persons were playing
computer games, and that the computer games she used were graphical and with more direct motor-visual control of the player character.
Since results similar to those of Johansson were obtained in the present study, Johansson’s results seem to generalise beyond all of the
three conditions mentioned.
The third meaning of jag (‘I’) found by Linderoth (2004)—that of
referring to oneself when playing the role of the game character—
was not found in the present study. The reason may be that the pre-
Perspective on actions and events 235
sent study investigated talk about events from the computer game
in a later situation, while Linderoth recorded talk during the computer game playing. It may also be that Linderoth’s third meaning
of jag is particular to children.
The referential ambiguity of first-person pronouns leads to
methodological difficulties. There is a problem as regards what
should count as an event that took place in the game world, compared to in the game playing situation. The use of jag (‘I’) cannot unproblematically be taken as a reference to the individual in the game
playing situation. Consider example 8.17. Here, the participant refers to a door in the game world, but the problem was something
that belonged to the playing situation. (The methodological solution, though not optimal, was to include these unclear cases in the
total count together with cases on the narrative level.)
(8.17) Referential ambiguity in the use of jag (participant 1):
när man väl kom till huset så va det bara liksom egentligen
once you got to the house it was basically a matter of
å (.) lägga sig försöka lägga sig å stänga dörren som jag hade
(.) going to bed try to go to bed and close the door that I
problem med å så
had trouble with
There seem, however, to be ways of separating the two meanings of
jag. In the use of jag (‘I’) to mean the individual, references are to (i)
mental states such as thoughts, (ii) the physical surrounding in the
playing situation, and (iii) managing and progressing in the game as
an abstract object. In the use of the other meaning of jag, the player
character, there are references to the game world and its objects,
places, and characters. These characteristics can work as guidelines,
but they do not completely eliminate the referential ambiguity of
jag.
No participant referred to the player character with the thirdperson pronoun, e.g., hon (‘she’) or a noun, such as ‘the woman’.
The use of a first-person pronoun does not seem to depend on
whether the player character is described in the game as a person
the player is supposed to be ‘playing’, in contrast to indicating that
236 Participating in a story
the player is ‘playing herself’. Even in this case, where the player
character was clearly described as a woman, all participants (including the male participants) referred to the player character with jag
rather than with a third-person pronoun or noun. The possibility
can be ruled out that the participants were not aware that the player
character was a woman, for instance, because of inattention to this
information as it was presented, because all participants included
this fact in their talk about the computer game.
Game Ego. The lack of motor activity for direct control does not
seem to have lessened the tendency of the participants to view
themselves as being ‘in the game’. Wilhelmsson’s (2001) Point of
Being seems to have been established, because participants used ‘I’
when talking about what happened in the computer game, and
never adopted a second- or third-person perspective. Thus, the results do not support Wilhelmsson’s idea of a need for a strong tactile motor/kinesthetic link in order for a Point of Being to arise.
The results support Wilhelmsson’s application of Lakoff’s Self and
Subject. Participants talked as if they were themselves present in the
game world. Thus, participants’ Subjects seem to have been projected into the player character’s Self.
Agent perspective switches between jag and man. A surprising result
was the frequent perspective switches between jag and man when
participants talked about events from the computer game (on average a switch around every 7 occurrences of jag or man). The most
frequent reason for switching from jag to man was to talk about
events in the game that would happen to any player. In this way,
the switch allowed the participants to be more general. The most
common reason for switching in the other direction, from man to
jag, was to tell about specifics of what the participant did in the
computer game. The use of jag does not entail that what is described
would happen to any player playing the game, or that it is something that the designer of the game had in mind as something a
player should do in order to succeed in the game. In this way, the
use of jag expresses that what is being described from the game is
something occurring because of an action—a conscious decision—on
the part of the player.
Perspective on actions and events 237
The shifts did not appear to be signalled in any way other than
lexical choice. They were not associated with any non-verbal features, such as pauses, prosody, or gesture.
That perspective switches occurred is however perhaps not
surprising considering other research that has found that perspective
switching in describing spatial scenes is frequent and sometimes can
be more effective than maintaining a consistent one, even considering the cognitive cost for the interlocutors (Tversky, Lee, & Mainwaring, 1999).
8.3
COMPARING THE TWO STUDIES OF
PERSPECTIVE EXPRESSIONS
Do the results of the two studies of expressed perspective provide a
unified picture? In comparing the match between how perspective
was expressed in gesture and speech, the following assumptions
were made, based on McNeill’s (1992) theory of perspective: Iconic
gesture observer perspective is associated with third-person singular
and plural pronouns, and nouns, as well as with the indefinite pronoun man (‘one’/‘you’) (since it expresses distance). This perspective is here called distant. Iconic gesture subjective perspective is
associated with first-person singular and first-person plural pronouns, because they signal closeness. This perspective is here called
close. Table 8.9 shows the dominant perspectives for each of the five
conditions.
It is only for expression of perspective on actions and events
from the computer game that gesture and speech reveal the same
picture. In the Game condition, both subjective/close and observer/distant perspectives were adopted, and these occurred in
similar proportions for both gesture and speech. For all the other
four conditions, there were discrepancies in that a difference was
found between subjective and observer perspectives for gesture, but
no difference for perspective expressed in speech, or vice versa. In
all these cases, it is possible that a match between gesture and speech
could have been obtained, had the number of participants been
greater.
238 Participating in a story
Table 8.9. Summary of dominant perspectives found in gesture and verbal expressions.
Source
Game
Story
Personal
Tasks
Nonparticipatory
Expression of perspective
Gesture
Speech
(observer/subjective)
(distant/close)
–
–
–
distant
–
close
subjective
–
–
distant
Note: A dash means that no perspective dominated the other, i.e., both occur in
about equal proportions.
8.4
CONCLUSION
The results of two analyses regarding which perspectives were
adopted by participants on past actions and events revealed differences between the five conditions: a computer game, a special nonparticipatory version of the computer game, a short story, personal
experience, and practical laboratory tasks.
Considering first the analysis of perspective as expressed in gesture, the results revealed that in talking about the practical laboratory tasks, participants adopted mainly a close, inside perspective
(an iconic gesture subjective viewpoint). In the other four conditions, both ‘outside’ and ‘inside’ perspectives were present (proportions of subjective viewpoint and observer viewpoint were about
equal).
The analysis of how perspective was revealed in speech resulted
in two groups. Talk about the computer game, personal experience,
and practical laboratory tasks featured mostly a close perspective,
while talk about the short story and the non-participatory version
of the computer game featured mostly a distant perspective. The
analysis of perspective as manifested in speech give the most detailed
results. These results show that the participatory story and its nonparticipatory version elicited different perspectives. Participants
Perspective on actions and events 239
adopted a perspective on events and actions from a participatory
story similar to events and actions from personal experience. Thus,
it seems that participation is the reason for adopting a personal perspective in the participatory story.
The implications of the results of the perspective study on cognition and the concept of participatory stories are discussed in Chapter 9.
9
CHAPTER 9
Conclusions and future research
I
closing chapter, first, the theoretical framework and the
empirical findings will be summarised. Then, the consequences of
the empirical results will be considered for cognition and for the
central concept in this book: participatory stories.
N THIS
9.1
SUMMARY
9.1.1 Summary of the theoretical framework
Part I, the theoretical framework, aimed at answering the question:
What are participatory stories? In order to arrive at an answer to this
question, the concepts of story and fiction were discussed and defined
in cognitive terms. A story was defined as a mental representation of
at least two chronologically related events, including an actual or intended state change by an agent. The content of a mental representation R is fiction to a cognitive system C if and only if C believes that R
should not be evaluated in relation to the real world. This laid the
foundation for a definition of story participation and participatory
stories, thus providing a viable classification and characteristic of
participatory stories in terms of cognition. The importance of a
cognitive perspective was framed in a weak and a strong version,
where according to the strong cognitive thesis it is necessary to include cognitive aspects in order to show what differentiates
participatory stories from non-participatory stories, and the weak
cognitive thesis, stating that it is fruitful and enlightening to account
241
242 Participating in a story
for cognition when characterising participatory stories. It was argued that the only plausible solution is to adopt the strong version
of the cognitive thesis. A definition of story participation was proposed as perceiving at least two different potential event sequences of a
story, and causing one of them to occur by controlling the actions of an
agent in the story. A participatory story was defined as a physical system separate from the audience allowing participation in a fictive story.
In contrast to the foci of earlier research, participatory stories were
defined without reference to the sense modalities or the type of actions that are involved when experiencing them.
9.1.2 Summary of empirical results
In the second part of the book, the line of reasoning from Part I was
followed, resulting in a series of empirical studies of cognition of
participatory stories. An explorative method of obtaining data
about cognition was developed in order to study differences between participatory stories and non-participatory stories. Participants were exposed to events from five sources with varying degrees
of participation, fictionality, and authenticity:27 an interactive
fiction computer game, a short story, personally experienced events,
practical laboratory tasks, and a special non-participatory version of
the computer game. Afterwards, participants were interviewed
about the events. Language data were analysed from three main
viewpoints: spatial cognition, memory qualities, and perspective on
memories of events and actions. The results of the analyses are given
here and they are further related to cognition in Section 9.2 below.
When analysing participants’ verbal descriptions of spatiality
from the participatory story, it was found that participants exclusively used a survey descriptive strategy (i.e., giving a description
from the above) using an extrinsic frame of reference (e.g., north,
south). This was true for both the computer game and the nonparticipatory version. A marked difference was found regarding
how complete, accurate, consistent, and integrated these spatial deAuthenticity here means how true the material and situations in the study were to
naturally occurring non-study settings, i.e., ecological validity.
27
Conclusions and future research 243
scriptions were. Participants who played the computer game revealed elaborate, relatively complete, accurate, and integrated descriptions of spatiality. In contrast, participants who read the nonparticipatory version of the computer game provided sparse and
incomplete spatial descriptions of the game world.
Exploring differences between memories of events from the five
sources regarding memory qualities, results showed that verbal transcripts of event descriptions from the five sources were rated
roughly equally by external judges. This prompted the conclusion
that the reality monitoring framework is suited to explain differences between memories of events with external and internal origin,
but not suited to explain differences between memories with varying degree of fictionality, as studied in this book.
The analysis of how perspective on actions and events was revealed through gesture showed that participants used roughly equal
proportions of subjective (close) and observer (distant) viewpoints
(except when talking about the practical laboratory tasks, where
primarily a subjective viewpoint was used). This suggests that participants sometimes viewed themselves as being outside the event,
and sometimes as taking part inside the event.
Results from analysis of how perspective was expressed in speech
revealed two main groups. In the computer game, personal experience, and laboratory tasks, participants used mainly an inside perspective using the first-person pronoun jag (‘I’), but the common use
of the indefinite pronoun man (‘one’/‘you’), especially when talking events and actions from the computer game, added distance to
the perspective. In the other two conditions, participants mainly
adopted an outside perspective by using third-person pronouns and
nouns. There was a frequent perspective switching between jag (‘I’)
and man (‘one’/‘you’) when participants talked about events from
the computer game. The most frequent reason for switching from
jag to man was to talk about events in the game that would happen
to any player. In this way, the switch allowed the participants to be
more general. The most common reason for switching in the other
direction, from man to jag, was to tell about specifics of what the
participant did in the computer game. In this way, the use of jag
244 Participating in a story
expresses that what is being described from the game is something
occurring because of an action—a conscious decision—on the part of
the player.
9.2
THE IMPLICATION OF THE RESULTS FOR
COGNITION
The results of the empirical studies in this book have implications
for both the cognitive processes directly studied here (spatial cognition, memory qualities, and perspective), but also by raising questions about cognition in general and suggesting future research.
9.2.1 Spatial cognition
The analysis of the players’ spatial descriptions showed that players
of interactive fiction (in contrast to readers of fiction) spontaneously
created spatial mental representations of the participatory story
world. This conclusion was reached because of the completeness,
accuracy, consistency, and integration revealed through the player’s
spatial descriptions. The reason for forming spatial mental representations is probably that they help navigation. People did not rely
only on information available in the immediate situation in order to
navigate, as would be expected from a view of cognition as basically
situated, but instead internalised a great portion of the spatial world
they experienced in the computer game.
Because spatial cognition was studied through language in this
book, only a coarse-grained analysis was possible. In a future study,
one could study, for instance, distortions in a better way using finergrained data, by having participants draw (or build by computer)
maps. This could be compared to known distortions of spatial cognition in real environments explored by navigation.
Another question concerns when spatial mental representations
are constructed. Are they formed directly at encoding or as a consequence of rehearsal (because they are accessed and used)? This could
be tested by interrupting participants at various points while they
Conclusions and future research 245
are using a participatory story and studying their mental representations.
In order to see whether spatial cognition depends on the presentation of space or whether the results from the present study are
general, one could carry out studies of spatial descriptions and spatial mental representations when the participatory story presents
the story world in another way, such as in a graphical threedimensional first-person view.
9.2.2 Memory qualities
Memories of events from a participatory story (textual interactive
fiction) did not differ in characteristics typical of memories of external events (such as perceptual and contextual details), compared
to memories of personally experienced, real events, along the dimensions of the reality monitoring framework. This lack of difference can mean at least two things regarding cognition; that memories from participatory stories are vivid and life-like as memories of
real events, or that the personally experienced events in the study
did not have the qualities typically associated with memories of external events. But regardless of interpretation, these two types of
memories seem to be similar in terms of their characteristic qualities. In that respect, the two types of memories would work in cognitively similar ways. However, since the participants had no trouble separating memories of fictional events from memories of real
events, there must be other cognitive processes at work—besides
those proposed in the reality monitoring framework—that enable
this separation.
In the reality monitoring framework, actual events are linked to
external events while imagined events are linked to internal events.
This may seem like a straightforward distinction. But what kind of
events are events from a participatory story? What does ‘actual’ and
‘imagined’ mean? There is a need to tease apart the external/internal
dimension from the fictional/factual dimension.
Related to the issue of memory qualities are field and observer
modes of remembering (Nigro & Neisser, 1983; Robinson & Swan-
246 Participating in a story
son, 1993; McIsaac & Eich, 2002). The results of analysis of the participant’s spatial descriptions—survey descriptions using an extrinsic
frame of reference—suggest that people may more often use an observer perspective when remembering events from a participatory
story (at least from a work of interactive fiction such as Anchorhead). Systematicity of modes of remembering of events from
participatory stories could be studied by explicitly asking participants (as in the studies of, e.g., McIsaac & Eich, 2002).
9.2.3 Gesture and cognition
According to McNeill (1992), the viewpoint expressed in iconic gesture when describing an event is a manifestation of the speaker’s
perspective (in cognition) on that event. Character viewpoint (called
subjective viewpoint in this book) indicates a psychological closeness
to the event while an observer viewpoint indicates a psychological
distance to the event. In contrast, the analysis of perspective as
manifested in gesture in the present book casts doubt on this view
because the analysis of perspective as manifested in gesture did not
correspond to gesture as manifested in speech. Perhaps the gesture
viewpoint is not a function of how close or distant the speaker is in
relation to the event, but it depends on the spatial configuration of
the event. When describing some events, it may suffice to use the
space immediately around the body in a one-to-one scale, which
would lead to a character viewpoint. Other events may require a
larger space; it is then compressed, resulting in an observer viewpoint. By manipulating the type of event, one could experimentally
study the effect on gesture in order to answer this question.
9.2.4 Self
It is perhaps the manifestation of perspective in speech (rather than
gesture) that gives the clearest picture of how participants seem to
think about the events and actions from the five sources. The feature of participation seems to shape thinking about events and actions from a participatory story into a form which is similar to
thinking about events from past personal experience and tasks per-
Conclusions and future research 247
formed in a laboratory. This conclusion is particularly persuasive
when contrasting the perspectives used about the participatory
story and the non-participatory version where the textual expressions of events in the sources are identical. Participants used the
pronoun jag (‘I’) when referring to the actions and events in the
participatory story (even though a reasonable choice of pronoun
would have been hon (‘she’), since the participants knew that the
participatory story was about a woman). Thus, the participants
viewed the actions and events as happening to the self rather than to
a character. What consequences may this have for cognition?
There might be differences concerning encoding of events that
involve the participant compared to more unconnected events that
happen in the story but are not brought about by the participant.
Actions that are enacted compared to read or watched have been
found to lead to better recall (Nilsson, 2000). One would expect to
find differences found in other research fields between actions that
are self-performed and actions that are not self-performed. For example, according to the self-generation effect (Symons & Johnson,
1997), memory performance should be superior for actions in a participatory story compared to a non-participatory story. This question would be of special interest when considering educational uses
of participatory stories.
A related but distinct question is whether more intense immersion leads to a better recall of the narrative (although there is little
to suggest that participatory stories generally have more intense
immersion than non-participatory stories—see Chapter 4).
Finally, the thinking about actions and events in a participatory
story may have consequences for research on autobiographical
memory, as studied within cognitive psychology. Autobiographical
memory is said to be unique because it is closely associated with a
sense of self:
In our approach this memory awareness or feeling state (the sense of
self [italics added] in the past) signals to a rememberer that the mental
representation it is associated with is in fact a memory of an experi-
248 Participating in a story
ence that actually occurred and is not a fantasy, dream, plan or some
other (experience-distant) mental construction. (Conway, 2002, p. 54)
What does ‘autobiographical’ in autobiographical memory mean? Is
it memory about episodes involving the individual herself, or also
episodes involving a participatory story agent? Simply put: Are
memories of events from participatory stories autobiographical? As
with the reality monitoring framework discussed above, research on
autobiographical memory does not seem to acknowledge the fictional status of events. Following the quotation above, we would
erroneously conclude that the audience (which relate to memories
using the self as shown in the perspective study), when remembering an event from a participatory story assumes that this memory is
of an event that actually occurred in real life and not only in an
imaginary story world. Earlier research has shown that autobiographical events differ neurologically from other kinds of events
(‘public events’ and other knowledge), as studied by brain imaging
techniques (Maguire, 2002). Research may answer the question of
whether events from participatory stories are autobiographical by
comparing neural activation of memories of participatory story
events compared to real world personal experience, ‘public events’,
and other knowledge.
9.2.5 Other cognitive aspects to be explored
In the present work, off-line cognition has been studied, that is, cognition taking place after people have experienced a participatory
story. A theme not explored in the present book is on-line cognition,
that is, what is going on cognitively while experiencing a participatory story. Here, a number of questions present themselves. Is a
different amount of attention required in comprehension? There
might be higher cognitive loads on the comprehender’s limited attentional capacity since her role is not confined to interpretation
only, but also to participation. Other differences may concern inferences found in comprehension of traditional narrative, for instance,
predictive inferences, on-line versus off-line inferences, and the time
Conclusions and future research 249
duration inferences are active (e.g., Trabasso & Suh, 1993). When
people comprehend traditional narratives, to some extent they generate predictive inferences about what is to happen next. Now, since
the user intervenes in what happens next in participatory stories,
predictive inferences may be affected. There could be more predictive inferences, they could be active for a longer time, or they could
be of a different kind. Another difference may concern monitoring,
how comprehenders keep track of the narrative characters’ knowledge and point of view, for instance, ‘who said what’ and ‘who
knows what’ in the story (see Graesser, Bowers, Olde, White, &
Person, 1999), since the player is controlling one of the characters.
Finally, how is the feeling of presence, or immersion, affected by being able to intervene in the narrative? The received view appears to
be that intervening leads to greater immersion, so that participatory
stories are more immersive than non-participatory stories (e.g., a
simulated three-dimensional world compared with a traditional
book). Such conclusions have only been supported by intuition. It
might just as well be the other way around; that a greater amount of
user intervention disturbs the feeling of immersion. Actually, some
studies suggest that in real classroom learning situations with multimedia stories, interaction hinders immersion (Plowman, 1998).
Empirical studies are needed in this area.
9.3
WHAT ARE PARTICIPATORY STORIES?
The implications of the empirical studies in Part I I of this book for
the concept of participatory stories as put forward in Part I will
now be discussed. Generally, the results of the empirical studies of
the book show that participatory stories (such as a work of interactive fiction) are things very different from non-participatory stories
(such as a short story). Let us look more closely at the concept of
participatory stories concerning three issues: spatiality, actions and
event sequences, and the modality-independence assumption. The
discussion ends with a consideration of the possibility of generalisa-
250 Participating in a story
tion from the specific empirical studies to participatory stories in
general.
9.3.1 Spatiality
The finding that players of interactive fiction created spatial mental
representations is not reflected in the definition of participatory
stories. However, as the definition gives only minimal requirements
of a participatory story, there is no demand that a participatory
story involves a spatial world at all. Nonetheless, the spatial mental
representations can be said to act as support for the decision part
(causing one event sequence to occur rather than another), involved
in all participatory stories, according to the proposed definition. If
decisions involve spatial movement of the agent in the story, spatial
mental representations give the basis for such decisions. It seems
that it is the element of participation that causes spatial mental representations to form, since readers of non-participatory stories did
not form these (see also Zwaan & van Oostendorp, 1993; Hakala,
1999). The likely reason for forming spatial mental representations
is to facilitate navigation. However, it cannot be ruled out that they
serve other purposes, such as general understanding of the participatory story, or that they have a purely aesthetic function.
Another point shown by the analysis of spatial cognition is that
even though a participatory story may seem fragmented at a glance,
it can induce a coherent spatial mental representation in the audience.
9.3.2 Actions and event sequences
What can the results of the empirical studies tell us about the part of
participatory stories which deals with causing events sequences to
occur, that is, action? From the accounts of events from the participatory story given by participants, it is not directly apparent that a
choice between event sequences had been performed while they
were involved in the participatory story. However, when participants talked about what happened and what was performed, it was
implied that out of the possible actions that could be taken, one was
Conclusions and future research 251
chosen. This could particularly be tied to the switch to and use of
the agent perspective jag (‘I’) (as opposed to man, ‘one’/‘you’).
The results from the study of perspective as expressed in speech
show a striking case of treating the story agent as an extension of
oneself. In all of the data where participants were talking about actions from the computer game, not a single reference was made to
the agent as agent (for instance, using third-person pronouns or
nouns). Participants talked about actions in the story world as if
they were carried out by themselves (or used an indefinite pronoun
to refer to players in general). This finding may even warrant a revision of the original definition of participatory stories, introducing
the notion that actions should be performed as if they were performed by oneself, and not via a story agent. However, this may
prove overly restrictive, since it after all does not seem to be a necessary feature of story participation.
The results from the reality monitoring memory quality analysis
are consistent with the view that participants related personally to
the actions in the participatory story. The memory qualities did not
differ between talk about events and actions from a participatory
story and real personal experience.
9.3.3 Modality-independence
In Chapter 4, it was argued that participatory stories should be
defined without reference to the sense modalities or specific actions
involved when experiencing them (called the modality-independence
assumption). In contrast, Aarseth (1997) and Wilhelmsson (2001),
emphasise the importance of physical, motor actions when describing the essence of participatory stories. Wilhelmsson also expresses
the view that the visual modality (and to some extent also the auditive modality) is especially tied to participatory stories. What do the
results of the empirical studies tell us about this independence assumption?
From the results of the studies of spatiality, it can be seen that
elaborate spatial mental representations can be formed with no visual perceptual input other than text.
252 Participating in a story
Regarding memory qualities, the results show that it does not
matter whether the source is purely textual or rich in sensory modality, when it comes to perceptual details of memories.
Results from the study of perspective in speech show two things
concerning modality independence. First, similar perspectives were
adopted when talking about the computer game, the laboratory
tasks, and real personal experience. This happened even though the
computer game was purely textual while the laboratory tasks and
prior personal events were experienced with multiple sense modalities. Second, neither a graphical, visual representation (other than
text) nor a direct motor-kinesthetic link is necessary in order to
think of oneself as ‘being in the game’—in Wilhelmsson’s (2001)
words establishing a Point of Being.
9.3.4 Generalisations
Since a participatory story of a certain kind was used in the empirical studies in this book (Anchorhead, which is an instance of interactive fiction), the question is to what extent the results can be generalised to other kinds of participatory stories. The possibilities for
generalisation may seem hopelessly remote, and all that one can say
anything about is the specific instance used in the study, and not the
general category. However, let us remember that this is nothing
unique to the studies of the present book. In any empirical study,
such as experimental studies, one uses a particular instance of something as stimuli but still generalises beyond those particular stimuli.
Thus, the issues become a matter of arguing for prototypicality of
the selected instance. In the present book, attempts were made to
see to that the chosen participatory story was a prototypical instance of participatory stories (according to the definition proposed
in Chapter 4).
Consequently, the findings of this book would be possible to
generalise concerning memory qualities and the perspective that
people adopt on events and actions. There may be some differences
concerning spatiality. Strategies for describing spatiality may differ
because of varying ways of presenting the spatiality of the participa-
Conclusions and future research 253
tory story world (e.g., a first-person viewpoint three-dimensional
presentation of some graphical participatory stories compared to
the textual presentation of Anchorhead). The use of the perspective
expressed by the Swedish indefinite pronoun man (‘one’/‘you’) may
not generalise to participatory stories other than those framed in a
game context (such as live role-playing and children’s pretence),
since this perspective was mainly used to talk about what one should
do in the game. Other than for these aspects, it is predicted that the
results of the studies in this book will generalise to participatory
stories in general.
REFERENCES
Aarseth, E. J. (1994). Nonlinearity and literary theory. In G. P. Landow
(Ed.), Hyper/text/theory (pp. 51–86). Baltimore & London: Johns Hopkins University Press.
Aarseth, E. J. (1997). Cybertext: Perspectives on ergodic literature. Baltimore,
MD: Johns Hopkins University Press.
Alexander, J. (1999). Screen play: audiovisual narrative and viewer interaction. Ph.D dissertation, Department of Cinema Studies, Stockholm
University, Stockholm.
Alonso-Quecuty, M. L. (1992). Deception detection and reality monitoring:
A new answer to an old question? In F. Lösel, D. Bender, & T. Bliesener
(Eds.), Psychology and law: International perspectives (pp. 228–332). Berlin: Walter de Gruyter.
Andersson, L-G. (1972). Man: ett pronomen. Gothenburg Papers in Theoretical Linguistics 15. Göteborg: Department of Linguistics, Göteborg
University, Sweden.
Aristotle. (1999). Poetics (S. H. Butcher, Trans.). Retrieved December 3,
2004, from http://www.gutenberg.org/dirs/etext99/poetc10.txt
(Original work published 350 B C )
Baddeley, A., Aggleton, J. P., & Conway, M. A. (2002). Episodic memory:
New directions in research. Oxford: Oxford University Press.
Barthes, R. (1996). Introduction to the structural analysis of narratives. In S.
Onega, & J. A. G. Landa (Eds.), Narratology (pp. 45–60). New York:
Longman.
Bartholomew, R. E. (1998). The Martian panic sixty years later. Skeptical
Inquirer, 22(6). Retrieved December 3, 2004, from
http://www.csicop.org/si/9811/martian.html
Bartlett, F. C. (1932). Remembering: A study in experimental and social psychology. London: Cambridge University Press.
255
256 Participating in a story
Bennardo, G. (2002). Map drawing in Tonga, Polynesia: Accessing mental
representations of space. Field Methods, 14(4), 390–417.
Blank, M. (1982). Deadline [Computer software]. Cambridge, MA: Infocom.
Block, N. (1980). Introduction: What is functionalism? In N. Block (Ed.),
Readings in philosophy of psychology (pp. 171–184). Cambridge, MA,
Harvard University Press.
van den Broek, P., Lorch, R. F. Jr., Linderholm, T., & Gustafson, M.
(2001). The effects of readers’ goals on inference generation and memory
for texts. Memory & Cognition, 29(8), 1081–1087.
Brockmole, J. R., & Wang, R. W. (2002). Switching between environmental
representations in memory. Cognition, 83, 295–316.
Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University
Press.
Buckles, M, A. (1985). Interactive fiction: The computer storygame ‘Adventure’. Ph.D dissertation, University of California, San Diego, CA.
Cassell, J., & Jenkins, H. (Eds.) (1998). From Barbie to Mortal Kombat: Gender and computer games. Cambridge, MA: MIT Press.
Castañeda, H.-N. (1989). Thinking, language, and experience. Minneapolis,
MN: University of Minnesota Press.
Chafe, W. (1994). Discourse, consciousness, and time. Chicago: University of
Chicago Press.
Chatman, S. (1993). Story and discourse: Narrative structure in fiction and
film. London: Cornell University Press.
Conway, M. A. (2002). Sensory-perceptual episodic memory and its context:
Autobiographical memory. In A. Baddeley, J. P. Aggleton, & M. A.
Conway (Eds.) Episodic memory: New directions in research (pp. 53–70).
Oxford: Oxford University Press.
Costanzo, W. V. (1986). Reading interactive fiction: implications of a new
literary genre. Educational Technology, 26(6), 31–35.
Crawford, C. (1993). Ga-ga over graphics. Interactive Entertainment Design,
7. Retrieved December 3, 2004, from
http://www.erasmatazz.com/library/JCGD_Volume_7/Ga-GaGraphics.html
Crawford, C. (1994). Representation versus depiction. Interactive Entertainment Design, 8. Retrieved December 3, 2004, from
http://www.erasmatazz.com/library/JCGD_Volume_8/Representatio
nDepiction.html
Davenport, G., & Murtaugh, M. (1997). Automatist storyteller systems and
the shifting sands of story. IBM Systems Journal, 36(3), 446–456.
References 257
Denis, M. & Denhière, G. (1990). Comprehension and recall of spatial descriptions. European Bulletin of Cognitive Psychology, 10, 115–143.
Dieberger, A. (1994). Navigation in textual virtual environments using a city
metaphor. Ph.D dissertation, Faculty of Technology and Sciences, Vienna University of Technology, Austria. Retrieved December 3, 2004,
from http://homepage.mac.com/juggle5/WORK/publications/thesis/
Thesis.html
Diwadkar, V. A., & McNamara, T. P. (1997). Viewpoint dependence in
scene recognition. Psychological Science, 8(4), 302–307.
Douglas, J. Y. (1992). Gaps, maps and perception: What hypertext readers
(don’t) do. Perforations, 3(1). Retrieved December 3, 2004, from
http://web.nwe.ufl.edu/~jdouglas/perforations.pdf
Doyle, A. C. (2000). The adventure of the red circle. Retrieved January 4,
2004, from ftp://ftp.sunet.se/pub/etext/gutenberg/etext00/rcrcl10.txt
(Original work published 1911)
Ekman, P., & Friesen, W. V. (1969). The repertoire of nonverbal behavioral
categories: origins, usage, and coding. Semiotica, 1, 49–98.
Foltz, P. W. (1996). Comprehension, coherence and strategies in hypertext
and linear text. In J.-F. Rouet, J. J. Levonen, A. P. Dillon, & R. J. Spiro
(Eds.), Hypertext and cognition. Hillsdale, NJ: Lawrence Erlbaum Associates.
Freundschuh, S. (2000). Micro- and macro-scale environments. In R. Kitchin
& S. Freundschuh (Eds.), Cognitive mapping: past, present and future (pp.
125–146). London: Routledge.
Furnham, A., De Siena, S., & Gunter, B. (2002). Children’s and adults’
recall of children’s news stories in both print and audio-visual presentation modalities. Applied Cognitive Psychology, 16, 191–210.
Gander, P. (1999). Two myths about immersion in new storytelling media.
Lund University Cognitive Studies, 80.
Gander, P. (2004). Spatial mental representations in interactive fiction: What
is particular about the interactive text? In S. Porhiel, & D. Klingler
(Eds.), L’Unité texte (pp. 96–124). Pleyben, France: Perspectives.
Gardner. H. (1987). The mind’s new science: A history of the cognitive revolution. New York: BasicBooks.
Gentry, M. S. (1998). Anchorhead: release 5, serial number 990206 [Computer
software]. Retrieved January 4, 2004, from
http://www.ifarchive.org/if-archive/games/zcode/anchor.z8
Graesser, A. C. (1981). Prose comprehension beyond the word. New York:
Springer-Verlag.
258 Participating in a story
Graesser, A. C., Bowers, C., Olde, B., White, K., & Person, N. K. (1999).
Who knows what? Propagation of knowledge among agents in a literary
storyworld. Poetics, 26(3), 143–175.
Graesser, A. C., Millis, K. K., & Zwaan, R. A. (1997). Discourse comprehension. Annual Review of Psychology, 48, 163–189.
Graesser, A. C., Swamer, S. S., & Hu, X. (1997). Quantitative discourse
psychology. Discourse Processes, 23, 229–263.
Guba, E. G. & Lincoln, Y. S. (1994). Competing paradigms in qualitative
research. In N. Denzin & Y.S. Lincoln (Eds.), Handbook of qualitative
research (pp. 105–117). Thousand Oaks, CA: Sage.
Hakala, C. M. (1999). Accessibility of spatial information in a situation
model. Discourse Processes, 27(3), 261–279.
Hansen, J. P., Andersen, A. W., & Roed, P. (1995). Eye-gaze control of
multimedia systems. In Y. Anazai, K. Ogawa, & H. Mori (Eds.), Symbiosis of human and artifact. Proceedings of the Sixth International Conference
on Human Computer Interaction, Tokyo, Japan (pp. 37–42). Amsterdam:
Elsevier.
Hernández-Fernaud, E., & Alonso-Quecuty, M. (1997). The cognitive interview and lie detection: A new magnifying glass for Sherlock Holmes?
Applied Cognitive Psychology, 11, 55–68.
Holmquist, L. E. (1997). Loose scripts and the influence engine: A model for
controlling multi-actor interactive narratives. Proceedings of Writing and
Computers 10, Brighton, UK.
Hutchins, E. (1995). Cognition in the wild. Cambridge, MA: MIT Press.
Johansson, B. (2000). ‘Kom och ät!’ ‘Jag ska bara dö först.’ Datorn i barns vardag. Göteborg, Sweden: Etnologiska Föreningen i Västsverige.
Johnson, M. K., Foley, M. A., Suengas, A. G., & Raye, C. L. (1988). Phenomenal characteristics of memory for perceived and imagined autobiographical events. Journal of Experimental Psychology: General, 117, 371–
376.
Johnson, M. K., Kahan, T. L., & Raye, C. (1984). Dreams and reality monitoring. Journal of Experimental Psychology: General, 113, 329–344.
Johnson, M. K., & Raye, C.L. (1981). Reality monitoring. Psychological
Review, 88, 67–85.
Joyce, M. (1987). Afternoon: a story [Computer software]. Cambridge, MA:
Eastgate.
Kant, I. (2004). The critique of pure reason (J. M. D. Meiklejohn, Trans.).
Retrieved December 3, 2004, from
http://www.gutenberg.org/dirs/etext03/cprrn10.txt (Original work
published 1787)
References 259
Kintsch, W. (1988). The use of knowledge in discourse processing: A construction-integration model. Psychological Review, 95, 363–394.
Kirk, R. (1974). Sentience and behaviour. Mind, 81, 43–60.
Kitchin, R. & Freundschuh, S. (2000). Cognitive mapping. In R. Kitchin &
S. Freundschuh (Eds.), Cognitive mapping: Past, present and future (pp. 1–
8). London: Routledge.
Labov, W., & Waletzky, J. (1967). Narrative analysis: Oral versions of personal experience. In J. Helm (Ed.) Essays on the verbal and visual arts (pp.
12–44). Seattle: University of Washington Press.
Lakoff, G. (1996). Sorry, I am not myself today: The metaphor system for
conceptualizing the self. In G. Fauconnier & E. Sweetser (Eds.), Spaces,
worlds, and grammar (pp. 91–123). Chicago and London: The University
of Chicago Press.
Landow, G. P. (1997). Hypertext 2.0: The convergence of contemporary critical
theory and technology. Baltimore, MD: Johns Hopkins University Press.
Lawrence, R. (1999). WinFrotz Version 2.32 R5.3 [Computer software].
Retrieved January 4, 2004, from
http://www.ifarchive.org/if-archive/infocom/interpreters/frotz/
WinFrotzR53.zip
Levelt, W. H. (1996). Perspective taking and ellipsis in spatial descriptions.
In P. Bloom, M. A. Peterson, L. Nadel, & M. F. Garrett (Eds.), Language and Space (pp. 77–109). Cambridge, MA: MIT Press.
Levinson, S. C. (2003). Space in language and cognition: Explorations in cognitive diversity. Cambridge: Cambridge University Press.
Li, P. & Gleitman, L. (2002). Turning the tables: Language and spatial reasoning. Cognition, 83, 265–294.
Linderoth, J. (2004). Datorspelandets Mening: Bortom idén om den interaktiva
illusionen. Ph.D dissertation, Department of Educational Sciences,
Göteborg University, Sweden.
Magliano, J. P., Dijkstra, K., & Zwaan, R. A. (1996). Generating predictive
inferences while viewing a movie. Discourse Processes, 22, 199–224.
Magliano, J. P., Miller, J., & Zwaan, R. A. (2001). Indexing space and time
in film understanding. Applied Cognitive Psychology, 15, 533–545.
Mandler, J. M., & Johnson, N. S. (1977). Remembrance of things parsed:
story structure and recall. Cognitive Psychology, 9, 111–151.
Mani, K. & Johnson-Laird, P. N. (1982). The mental representation of spatial descriptions. Memory & Cognition, 10, 181–187.
Marsh, E. J., Meade, M. L., & Roediger, H. L. (2003). Learning facts from
fiction. Journal of Memory & Language, 49, 519–536.
260 Participating in a story
Mark, D. M., Freksa, C., Hirtle, S. C., Lloyd, R., & Tversky, B. (1999).
Cognitive models of geographical space. International Journal of Geographical Information Science, 13(8), 747–774.
Mateas, M. (1997). An Oz-centric review of interactive drama and believable
agents. Technical Report CMU-CS-97-156, School of Computer Science,
Carnegie Mellon University, Pittsburgh, PA. Retrieved December 3,
2004, from
http://www-cgi.cs.cmu.edu/afs/cs.cmu.edu/project/oz/web/papers/
CMU-CS-97-156.html
McIsaac, H. K., & Eich, E. (2002). Vantage point in episodic memory. Psychonomic Bulletin & Review, 9, 146–150.
McNeill, D. (1992). Hand and mind: What gestures reveal about thought.
Chicago: Chicago University Press.
Moulthrop, S. (1991). Toward a paradigm for reading hypertexts: Making
nothing happen in hypermedia fiction. In E. Berk, & J. Devlin (Eds.),
Hypertext/hypermedia handbook (pp. 65–78). New York: McGrawHill/Intertext.
Mullin, E. (1999a). X Y Z Z Y Awards: 1998 winners. Retrieved January 4, 2004,
from http://www.xyzzynews.com/98winners.html
Mullin, E. (1999b). X Y Z Z Y news: 1998 X Y Z Z Y Awards nominees. Retrieved
January 4, 2004, from
http://www.xyzzynews.com/98nominees.html
Murray, J. H. (1997). Hamlet on the holodeck: The future of narrative in cyberspace. New York: The Free Press.
Nichols, S., & Stich, S. (2000). A cognitive theory of pretense. Cognition,
74, 115–147.
Niedźwieńska, A. (2003). Distortion of autobiographical memories. Applied
Cognitive Psychology, 17, 81–91.
Nigro, G., & Neisser, U. (1983). Point of view in personal memories. Cognitive Psychology, 15, 467–482.
Nilsson, L.-G. (2000). Remembering actions and words. In E. Tulving, & F.
I. M. Craik (Eds.), The Oxford handbook of memory (pp. 137–148). Oxford, UK: Oxford University Press.
Norell, P. (1995). ‘Answers to that question were now being sought’; English translations of the Swedish indefinite pronoun man in fiction and
non-fiction texts. In G. Melchers, & B. Warren (Eds.), Studies in anglistics (pp. 191–200). Stockholm: Almqvist & Wiksell International.
Norrick, N. R. (2000). Conversational narrative: storytelling in everyday talk.
Amsterdam: John Benjamins.
References 261
Paivio, A. (1991). Dual coding theory: retrospect and current status. Canadian Journal of Psychology, 45(3), 255–287.
Parsons, T. (1980). Nonexistent objects. New Haven, CT: Yale University
Press.
Pazhitnov, A. (1985). Tetris [Computer software].
Perfetti, C. A. (1996): Text and hypertext. In J.-F. Rouet, J. J. Levonen, A.
Dillon, & R. J. Spiro (Eds): Hypertext and Cognition (pp. 157–163).
Mahwah, NJ: Lawrence Erlbaum.
Perrig, W. & Kintsch, W. (1985). Propositional and situational representations of text. Journal of Memory and Language, 24, 503–518.
Plowman, L. (1996a). Narrative, linearity and interactivity: Making sense of
interactive multimedia. British Journal of Educational Technology, 27, (2),
92–105.
Plowman, L. (1996b). Towards an understanding of interactive multimedia
‘literacy’: Mapping the space. Paper presented at Literacy & Culture, the
33rd annual conference of the United Kingdom Reading Association,
July 1996.
Plowman, L. (1998). Getting side-tracked: Cognitive overload, narrative,
and interactive learning environments. To appear in Virtual Learning
Environments and the Role of the Teacher, Proceedings of
U N E S C O /Open University International Colloquium, Milton Keynes,
UK, April 1997.
Potts, G. R., St. John, M. F., & Kirson, D. (1989). Incorporating new information into existing world knowledge. Cognitive Psychology, 21, 303–
333.
Prentice, D. A., & Gerrig, R. J. (1999). Exploring the boundary between
fiction and reality. In S. Chaiken, & Y. Trope (Eds.), Dual process theories in social psychology (pp. 529–546). New York: Guilford Press.
Prince, G. (1983). Narrative pragmatics, message, and point. Poetics, 12, 527–
536.
Prince, G. (1988). A dictionary of narratology. Lincoln, NE: University of
Nebraska Press.
Propp, V. J. (1968). Morphology of the folktale (2nd ed.) (L. Scott, Trans.).
Austin, TX: University of Texas Press. (Original work published 1928)
Rapaport, W. J., & Shapiro, S. C. (1995). Cognition and fiction. In J. F.
Duchan, G. A. Bruder, & L. E. Hewitt (Eds.), Deixis in narrative: a cognitive science approach (pp. 107–128). Hillsdale, NJ: Lawrence Erlbaum.
262 Participating in a story
Risden, K. (1997). Can theories of traditional narrative understanding inform the design of hypermedia stories? Paper presented at M E N O’s
Workshop on Narrative and Hypermedia. Retrieved October 14, 1998
from http://www-iet.open.ac.uk/iet/meno/HT97/kirsten.html
Robinson, J.A., & Swanson, J.A. (1993). Field and observer modes of remembering. Memory, 1, 169–184.
Ryan, M.-L. (1997). Interactive drama: Narrativity in a highly interactive
environment. Modern Fiction Studies, 43(3), 677–707.
Ryan, M.-L. (1999). Cyberspace textuality: computer technology and literary
theory. Bloomington and Indianapolis: Indiana University Press.
Schank, R. C. (1990). Tell me a story: A new look at real and artificial memory. New York: Scribner.
Schank, R. C. & Abelson, R. (1977). Scripts, plans, goals, and understanding.
Hillsdale, NJ: Earlbaum Assoc.
Schooler, J. W., Gerhard, D. & Loftus, E. F. (1986). Qualities of the unreal.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 12,
171–181.
Shneiderman, B., Kreitzberg, C, & Berk, E. (1991). Editing to structure a
reader’s experience. In E. Berk & J. Devlin (Eds.), Hypertext/Hypermedia
Handbook (pp. 143–164). New York: McGraw-Hill/Intertext.
Sloane, S. (1991). Interactive fiction, virtual realities, and the reading-writing
relationship. Ph.D dissertation, Ohio State University.
Sørensen, E. (1998). Computerspil: virekelighed eller fiktion? Nordisk Psykologi, 50(2), 135–150.
Sporer, S. L. (1997). The less travelled road to truth: verbal cues in deception
detection in accounts of fabricated and self-experienced events. Applied
Cognitive Psychology, 11(5), 373–397.
Strömwall, L. A., Bengtsson, L., Leander. L., & Granhag, P. A. (2004). Assessing children’s statements: the impact of a repeated experience on
C B C A and R M ratings. Applied Cognitive Psychology, 18, 653–668.
Suengas, A. G., & Johnson, M. K. (1988). Qualitative effects of rehearshal
on memories for percieved and imaginded complex events. Journal of
Experimental Psychology: General, 117, 377–389.
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect in memory: a meta-analysis. Psychological Bulletin, 121, 371–394.
Talmy, L. (1996). Fictive motion in language and ‘ception’. In P. Bloom, M.
A. Peterson, L. Nadel, & M. F. Garrett, (Eds.), Language and space (pp.
211–276). Cambridge, MA: MIT Press.
References 263
Taylor, H. A., & Tversky, B. (1992). Spatial mental models derived from
survey and route descriptions. Journal of Memory and Language, 31, 261–
292.
Taylor, H. A., & Tversky, B. (1996). Perspective in spatial descriptions.
Journal of Memory and Language, 35, 371–391.
Tolman, E. C. (1948). Cognitive maps in rats and men. The Psychological
Review, 55(4), 189–208.
Trabasso, T. & Suh, S. (1993). Understanding text: Achieving explanatory
coherence through on-line inferences and mental operations in working
memory. Discourse Processes, 16, 3–34.
Trabasso, T., & Magliano, J. P. (1996). Conscious understanding during
comprehension. Discourse Processes, 21, 255–287.
Tromp, J. G. (1993). Results of two surveys about spatial perception and
navigation of a text-based spatial interface. Retrieved December 3, 2004,
from http://www.cse.dmu.ac.uk/~cph/VR/JolaPaper/jola.html
Turkle, S. (1995). Life on the screen: Identity in the age of the Internet. New
York: Simon & Schuster.
Tversky, B. (1981). Distortions in memory for maps. Cognitive Psychology,
13, 407–433.
Tversky, B. (2000). Levels and structure of spatial knowledge. In R. Kitchin
& S. Freundschuh (Eds.), Cognitive mapping: Past, present and future (pp.
24–43). London: Routledge.
Tversky, B., Lee, P., & Mainwaring, S. (1999). Why do speakers mix perspectives?. Spatial Cognition and Computation, 1, 399–412.
Tversky, B. & Marsh, E. J. (2000). Biased retellings of events yield biased
memories. Cognitive Psychology, 40(1), 1–38.
Van Inwagen, P. (1977). Creatures of fiction. American Philosophical Quarterly, 14, 299–308
Vrij, A. (2000). Detecting lies and deceit: The psychology of lying and the implications for professional practice. Chichester: John Wiley & Sons Ltd.
Vrij, A., Akehurst, L., Soukara, S., & Bull, R. (2004). Detecting deceit via
analyses of verbal and nonverbal behavior in children and adults. Human Communication Research, 30(1), 8–41.
Vrij, A., Edward, K., Roberts, K. P., & Bull, R. (2000). Detecting deceit via
analysis of verbal and nonverbal behavior. Journal of Nonverbal Behavior, 24(4) 239–263.
Vrij, A., Evans, H., Akehurst, L., & Mann, S. (2004). Rapid judgements in
assessing verbal and nonverbal cues: Their potential for deception researchers and lie detection. Applied Cognitive Psychology, 18, 283–296.
264 Participating in a story
Wenger, M. J., & Payne, D. G. (1996). Comprehension and retention of
nonlinear text: considerations of working memory and materialappropriate processing. American Journal of Psychology, 109(1), 93–130.
Whitaker, J. C. (Ed.) (2001). Interactive television demystified. New York:
McGraw-Hill.
Wilhelmsson, U. (2001). Enacting the point of being: Computer games, interaction and film theory: affordances and constraints, metaphorical concepts
and experientialist cognition observed through the environment in computer
games. Ph.D dissertation, Department of Film and Media Studies, Copenhagen University.
Wilson, M. (2002). Six views of embodied cognition. Psychonomic Bulletin
and Review, 9, 625–636.
Wilson, S. G., Rinck, M., McNamara, T., Bower, G. H., & Morrow, D. G.
(1993). Mental models and narrative comprehension: Some qualifications. Journal of Memory and Learning, 32, 141–154.
Zacks, J. M., & Tversky, B. (2001). Event structure in perception and conception. Psychological Bulletin, 127(1), 3–21.
Zwaan, R.A., & van Oostendorp, H. (1993) Do readers construct spatial
representations in naturalistic story comprehension? Discourse Processes,
16, 125–143.
APPENDIX
Excerpt from log file transcript generated
while using the participatory story
Anchorhead (participant 4)
ANCHORHEAD
[Press ‘R’ to restore; any other key to begin]
November, 1997.
You take a deep breath of salty air as the first raindrops begin to spatter the
pavement, and the swollen, slate-colored clouds that blanket the sky mutter ominous
portents amongst themselves over the little coastal town of Anchorhead.
Squinting up into the glowering storm, you wonder how everything managed to
happen so fast. The strange phone call over a month ago, from a lawyer claiming to
represent the estate of some distant branch of Michael’s family, was bewildering
enough in itself... but then the sudden whirlwind of planning and decisions, legal
details and travel arrangements, the packing up and shipping away of your entire
home, your entire life...
Now suddenly here you are, after driving for the past two days straight, over a
thousand miles away from the familiar warmth of Texas, getting ready to move into
265
266 Participating in a story
the ancestral mansion of a clan of relatives so far removed that not even Michael has
ever heard of them. And you’ve only been married since June and none of this was
any of your idea in the first place, and already it’s starting to rain.
These days, you often find yourself feeling confused and uprooted.
You shake yourself and force the melancholy thoughts from your head, trying to
focus on the errand at hand. You’re to meet with the real estate agent and pick up
the keys to your new house while Michael runs across town to take care of some
paperwork at the university. He’ll be back to pick you up in a few minutes, and then
the two of you can begin the long, precarious process of settling in.
A sullen belch emanates from the clouds, and the rain starts coming down harder -fat, cold drops smacking loudly against the cobblestones. Shouldn’t it be snowing in
New England at this time of year? With a sigh, you open your umbrella.
Welcome to Anchorhead...
ANCHORHEAD
An interactive gothic by Michael S. Gentry
(Type HELP or ABOUT for some useful information.)
Release 5 / Serial number 990206 / Inform v6.15 Library 6/7
Outside the Real Estate Office
A grim little cul-de-sac, tucked away in a corner of the claustrophobic tangle of
narrow, twisting avenues that largely constitute the older portion of Anchorhead.
Like most of the streets in this city, it is ancient, shadowy, and leads essentially
nowhere. The lane ends here at the real estate agent’s office, which lies to the east,
and winds its way back toward the center of town to the west. A narrow, garbagechoked alley opens to the southeast.
>enter office
(opening the real estate office door first)
It seems to be locked.
>examine door
Appendix 267
The door has a glass front with the name of the real estate company -- Benson &
Brackhurst -- stenciled across it. The blinds are drawn, the lights are off inside, and
no one appears to be home. Odd, since the agent knew you were coming today.
>exa window
That’s not a verb I recognise.
>examine window
You can’t see any such thing.
>l
Outside the Real Estate Office
A grim little cul-de-sac, tucked away in a corner of the claustrophobic tangle of
narrow, twisting avenues that largely constitute the older portion of Anchorhead.
Like most of the streets in this city, it is ancient, shadowy, and leads essentially
nowhere. The lane ends here at the real estate agent’s office, which lies to the east,
and winds its way back toward the center of town to the west. A narrow, garbagechoked alley opens to the southeast.
In the distance, you can hear the lonesome keening of a train whistle drifting on the
wind.
>enter alley
Alley
This narrow aperture between two buildings is nearly blocked with piles of rotting
cardboard boxes and overstuffed garbage cans. Ugly, half-crumbling brick walls to
either side totter oppressively over you. The alley ends here at a tall, wooden fence.
High up on the wall of the northern building there is a narrow, transom-style
window.
>examine boxes
The boxes are filthy, slimy and soaked with rain.
>open boxes
Open or closed, there’s very little difference; it’s all one soggy mass of rotted
cardboard.
>search boxes
The piles of cardboard boxes are empty.
>examine cans
The metal garbage cans are stuffed to overflowing with slowly decomposing refuse.
268 Participating in a story
>examine refuse
The metal garbage cans are stuffed to overflowing with slowly decomposing refuse.
>search refuse
You are not about to go digging through garbage.
>dig garbage
Digging would achieve nothing here.
>l
Alley
This narrow aperture between two buildings is nearly blocked with piles of rotting
cardboard boxes and overstuffed garbage cans. Ugly, half-crumbling brick walls to
either side totter oppressively over you. The alley ends here at a tall, wooden fence.
High up on the wall of the northern building there is a narrow, transom-style
window.
>exa walls
That’s not a verb I recognise.
>examine walls
It’s just an ordinary-looking wall.
>search wall
You find nothing of interest.
>examine fence
One of the boards seems to be loose down at the bottom; you could probably just
squeeze through.
>squeeze through fence
(closing the umbrella first)
Dropping to your hands and knees, you wriggle underneath the loose board and
scramble down a muddy slope.
Narrow Beach
This narrow strip of beach is tucked away between two outcroppings in the
predominantly rocky shoreline, accessible only from a steep, muddy slope to the
west. The sand is filthy and strewn with rocks, seaweed, litter and other bits of
storm-tossed detritus.
Appendix 269
Near the bottom of the slope, a sewage outflow pipe juts out over the beach, about
three feet above the ground. A thin stream of acrid-smelling sewer water trickles out
over the lip of the pipe, forming a puddle in the sand.
You are getting wet.
>i
You are wearing your wedding ring, your trenchcoat and your clothes; in addition,
you have in your hand your umbrella.
The pockets of your trenchcoat are empty.
You are getting wet.
>exa ring
That’s not a verb I recognise.
>look ring
You can’t see any such thing.
>look at ring
Staring at your wedding ring like that tends to make you sentimental.
A sudden gust of wind blows a cold spray of rain into your face.
You are getting wet.
>examine ring
Staring at your wedding ring like that tends to make you sentimental.
Another wave crashes against the rocks, sending a cloud of spray into the air.
You are getting wet.
>open umbrella
You open the umbrella.
>l
Narrow Beach
This narrow strip of beach is tucked away between two outcroppings in the
predominantly rocky shoreline, accessible only from a steep, muddy slope to the
west. The sand is filthy and strewn with rocks, seaweed, litter and other bits of
storm-tossed detritus.
270 Participating in a story
Near the bottom of the slope, a sewage outflow pipe juts out over the beach, about
three feet above the ground. A thin stream of acrid-smelling sewer water trickles out
over the lip of the pipe, forming a puddle in the sand.
>look pipe
You can’t see any such thing.
>look at pipe
The concrete pipe is about a foot and a half in diameter. Looking in, you can only
see about three feet before the inner walls of the pipe disappear into blackness.
>enter pipe
You don’t really feel like crawling head first into a smelly, filthy, pitch black sewer
pipe.
>close umbrella
You close the umbrella.
You are getting wet.
>enter pipe
You don’t really feel like crawling head first into a smelly, filthy, pitch black sewer
pipe.
The clouds overhead mutter restlessly to themselves.
You are now, sadly, entirely soaked.
>laugh
That’s not a verb I recognise.
>enter pipe
You don’t really feel like crawling head first into a smelly, filthy, pitch black sewer
pipe.
>search pipe
You can only see a few feet into the pipe before the interior is lost in darkness.
>i
You are wearing your wedding ring, your trenchcoat and your clothes; in addition,
you have in your hand your umbrella.
The pockets of your trenchcoat are empty.
>l
Appendix 271
Narrow Beach
This narrow strip of beach is tucked away between two outcroppings in the
predominantly rocky shoreline, accessible only from a steep, muddy slope to the
west. The sand is filthy and strewn with rocks, seaweed, litter and other bits of
storm-tossed detritus.
Near the bottom of the slope, a sewage outflow pipe juts out over the beach, about
three feet above the ground. A thin stream of acrid-smelling sewer water trickles out
over the lip of the pipe, forming a puddle in the sand.
Another wave crashes against the rocks, sending a cloud of spray into the air.
>look at water
Which do you mean, the sea, the rain or the sewer water?
>look at sewer water
The water coming from the pipe is foul-smelling and brown.
>taste sewer water
You must be joking.
>yep
That’s not a verb I recognise.
>taste sea water
You’ve no wish to drink from these dismal waters.
>l
Narrow Beach
This narrow strip of beach is tucked away between two outcroppings in the
predominantly rocky shoreline, accessible only from a steep, muddy slope to the
west. The sand is filthy and strewn with rocks, seaweed, litter and other bits of
storm-tossed detritus.
Near the bottom of the slope, a sewage outflow pipe juts out over the beach, about
three feet above the ground. A thin stream of acrid-smelling sewer water trickles out
over the lip of the pipe, forming a puddle in the sand.
>look at sea
The sea is the color of old pewter, surging and chopping restlessly beneath the
clouds.
>enter sea
272 Participating in a story
The water doesn’t look inviting.
>look at rain
It’s coming down in uneven spatters, broken by intermittent gusts. Most of the time
you enjoy such weather, but this has to be the most thoroughly depressing rain
you’ve ever experienced in your life.
>l
Narrow Beach
This narrow strip of beach is tucked away between two outcroppings in the
predominantly rocky shoreline, accessible only from a steep, muddy slope to the
west. The sand is filthy and strewn with rocks, seaweed, litter and other bits of
storm-tossed detritus.
Near the bottom of the slope, a sewage outflow pipe juts out over the beach, about
three feet above the ground. A thin stream of acrid-smelling sewer water trickles out
over the lip of the pipe, forming a puddle in the sand.
>look at puddle
The water coming from the pipe is foul-smelling and brown.
>search puddle
You find nothing of interest.
Another wave crashes against the rocks, sending a cloud of spray into the air.
The rain slackens off momentarily to a weak drizzle, then returns afresh in a brief,
freezing downpour.
>open umbrella
You open the umbrella.
>l
Narrow Beach
This narrow strip of beach is tucked away between two outcroppings in the
predominantly rocky shoreline, accessible only from a steep, muddy slope to the
west. The sand is filthy and strewn with rocks, seaweed, litter and other bits of
storm-tossed detritus.
Near the bottom of the slope, a sewage outflow pipe juts out over the beach, about
three feet above the ground. A thin stream of acrid-smelling sewer water trickles out
over the lip of the pipe, forming a puddle in the sand.
Appendix 273
>look at sand
There is nothing of any worth or interest here, just drifts of trash.
>put arm in pipe
You can’t see any such thing.
>i
You are wearing your wedding ring, your trenchcoat and your clothes; in addition,
you have in your hand your umbrella.
The pockets of your trenchcoat are empty.
>put ring in puddle
(first taking it off)
You slip the plain gold band from your finger. Usually you feel almost naked
without your ring on, although sometimes you do find yourself taking it off briefly,
just to read the inscription along the inside.
That can’t contain things.
>read inscription
Engraved along the inside are Michael’s and your initials, along with your wedding
date -- June 28th.